From 6b0d80507c9726c2a59afe5dfa87d6c93cc49145 Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Tue, 6 Jun 2023 09:47:32 +0000
Subject: [PATCH 1/9] Add datatypes for Larch XAS tools

---
 .../config/sample/datatypes_conf.xml.sample   |   9 +
 lib/galaxy/datatypes/larch.py                 | 204 ++++++++++++++++++
 lib/galaxy/datatypes/test/bad_atoms.inp       |  44 ++++
 lib/galaxy/datatypes/test/bad_potentials.inp  |  24 +++
 lib/galaxy/datatypes/test/no_pymatgen.inp     |  51 +++++
 lib/galaxy/datatypes/test/test.inp            |  52 +++++
 lib/galaxy/datatypes/test/test.prj            | Bin 0 -> 5372 bytes
 7 files changed, 384 insertions(+)
 create mode 100644 lib/galaxy/datatypes/larch.py
 create mode 100644 lib/galaxy/datatypes/test/bad_atoms.inp
 create mode 100644 lib/galaxy/datatypes/test/bad_potentials.inp
 create mode 100644 lib/galaxy/datatypes/test/no_pymatgen.inp
 create mode 100644 lib/galaxy/datatypes/test/test.inp
 create mode 100644 lib/galaxy/datatypes/test/test.prj

diff --git a/lib/galaxy/config/sample/datatypes_conf.xml.sample b/lib/galaxy/config/sample/datatypes_conf.xml.sample
index 49aab75ddbb..226251828a6 100644
--- a/lib/galaxy/config/sample/datatypes_conf.xml.sample
+++ b/lib/galaxy/config/sample/datatypes_conf.xml.sample
@@ -946,6 +946,13 @@
     <datatype extension="castep" type="galaxy.datatypes.text:Castep" display_in_upload="true"/>
     <datatype extension="param" type="galaxy.datatypes.text:Param" display_in_upload="true"/>
     <datatype extension="den_fmt" type="galaxy.datatypes.text:FormattedDensity" display_in_upload="true"/>
+    <!-- Larch types -->
+    <datatype extension="prj" type="galaxy.datatypes.larch:AthenaProject" display_in_upload="true" description="Athena project file."/>
+    <datatype extension="inp" type="galaxy.datatypes.larch:FEFFInput" display_in_upload="true" description="FEFF input file."/>
+    <datatype extension="sp" type="galaxy.datatypes.tabular:CSV" subclass="true" description="CSV representing selected FEFF paths."/>
+    <datatype extension="gds" type="galaxy.datatypes.tabular:CSV" subclass="true" description="CSV representing GDS parameters."/>
+    <datatype extension="feff" type="galaxy.datatypes.tabular:CSV" subclass="true" description="CSV representing summary of FEFF paths."/>
+    <datatype extension="feffit" type="galaxy.datatypes.data:Text" subclass="true" description="Report of Artemis FEFFIT."/>
     <!-- ECOLOGY types -->
     <datatype extension="bil" type="galaxy.datatypes.binary:Binary" mimetype="application/octet-stream" display_in_upload="true" subclass="true" description="ENVI file with band interleave by line (BIL) format"/>
     <datatype extension="hdr" type="galaxy.datatypes.data:Text" mimetype="text/plain" display_in_upload="true" subclass="true" description="ENVI metadata header file"/>
@@ -1204,6 +1211,8 @@
     <sniffer type="galaxy.datatypes.media:Mpg"/>
     <sniffer type="galaxy.datatypes.speech:TextGrid" />
     <sniffer type="galaxy.datatypes.speech:BPF" />
+    <sniffer type="galaxy.datatypes.larch:FEFFInput" />
+    <sniffer type="galaxy.datatypes.larch:AthenaProject" />
     <sniffer type="galaxy.datatypes.text:Castep" />
     <sniffer type="galaxy.datatypes.text:CTLresult"/>
     <sniffer type="galaxy.datatypes.text:FormattedDensity" />
diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
new file mode 100644
index 00000000000..80f7fb358c0
--- /dev/null
+++ b/lib/galaxy/datatypes/larch.py
@@ -0,0 +1,204 @@
+from typing import TYPE_CHECKING
+
+from galaxy.datatypes.data import (
+    get_file_peek,
+    Text,
+)
+from galaxy.datatypes.metadata import MetadataElement
+from galaxy.datatypes.sniff import (
+    build_sniff_from_prefix,
+    FilePrefix,
+    get_headers,
+)
+
+if TYPE_CHECKING:
+    from galaxy.model import DatasetInstance
+
+
+@build_sniff_from_prefix
+class AthenaProject(Text):
+    """
+    Athena project format
+    """
+
+    file_ext = "prj"
+    compressed = True
+    compressed_format = "gzip"
+
+    MetadataElement(
+        name="atsym",
+        desc="Atom symbol",
+        readonly=True,
+        visible=True,
+    )
+    MetadataElement(
+        name="bkg_e0",
+        desc="Edge energy (eV)",
+        readonly=True,
+        visible=True,
+    )
+    MetadataElement(
+        name="edge",
+        desc="Edge",
+        readonly=True,
+        visible=True,
+    )
+    MetadataElement(
+        name="npts",
+        desc="Number of points",
+        readonly=True,
+        visible=True,
+    )
+    MetadataElement(
+        name="xmax",
+        desc="Maximum energy (eV)",
+        readonly=True,
+        visible=True,
+    )
+    MetadataElement(
+        name="xmin",
+        desc="Minimum energy (eV)",
+        readonly=True,
+        visible=True,
+    )
+
+    def sniff_prefix(self, file_prefix: FilePrefix) -> bool:
+        """
+        Try to guess if the file is an Athena project file.
+
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('test.prj')
+        >>> AthenaProject().sniff(fname)
+        True
+        """
+
+        return file_prefix.startswith("# Athena project file")
+
+    def set_meta(self, dataset: "DatasetInstance", overwrite: bool = True, **kwd) -> None:
+        """
+        Extract metadata from @args
+        """
+        headers = get_headers(dataset.file_name, sep=" = ", count=3, comment_designator="#")
+        for header in headers:
+            if header[0] == "@args":
+                args = header[1][1:-2].split(",")
+                break
+
+        index = args.index("'atsym'")
+        dataset.metadata.atsym = args[index + 1][1:-1]
+        index = args.index("'bkg_e0'")
+        dataset.metadata.bkg_e0 = args[index + 1][1:-1]
+        index = args.index("'edge'")
+        dataset.metadata.edge = args[index + 1][1:-1]
+        index = args.index("'npts'")
+        dataset.metadata.npts = args[index + 1][1:-1]
+        index = args.index("'xmax'")
+        dataset.metadata.xmax = args[index + 1][1:-1]
+        index = args.index("'xmin'")
+        dataset.metadata.xmin = args[index + 1][1:-1]
+
+    def set_peek(self, dataset: "DatasetInstance", **kwd) -> None:
+        if not dataset.dataset.purged:
+            dataset.peek = get_file_peek(dataset.file_name)
+            dataset.info = (
+                f"atsym: {dataset.metadata.atsym}\n"
+                f"bkg_e0: {dataset.metadata.bkg_e0}\n"
+                f"edge: {dataset.metadata.edge}\n"
+                f"npts: {dataset.metadata.npts}\n"
+                f"xmax: {dataset.metadata.xmax}\n"
+                f"xmin: {dataset.metadata.xmin}"
+            )
+            dataset.blurb = f"Athena project file of {dataset.metadata.atsym} {dataset.metadata.edge} edge"
+
+        else:
+            dataset.peek = "file does not exist"
+            dataset.blurb = "file purged from disk"
+
+
+@build_sniff_from_prefix
+class FEFFInput(Text):
+    """
+    FEFF input format
+    """
+
+    file_ext = "inp"
+
+    MetadataElement(
+        name="title_block",
+        desc="Title block",
+        readonly=True,
+        visible=True,
+    )
+
+    def sniff_prefix(self, file_prefix: FilePrefix) -> bool:
+        """
+        Try to guess if the file is an FEFF input file.
+
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('test.inp')
+        >>> FEFFInput().sniff(fname)
+        True
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('no_pymatgen.inp')
+        >>> FEFFInput().sniff(fname)
+        True
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('bad_atoms.inp')
+        >>> FEFFInput().sniff(fname)
+        False
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('bad_potential.inp')
+        >>> FEFFInput().sniff(fname)
+        False
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('Si.cif')
+        >>> FEFFInput().sniff(fname)
+        False
+        """
+        # pymatgen marks generated FEFF inputs, but user might also upload from another source
+        if file_prefix.startswith("* This FEFF.inp file generated by pymatgen"):
+            return True
+
+        generator = file_prefix.line_iterator()
+        line = next(generator)
+        while line is not None:
+            if line == "POTENTIALS":
+                line = next(generator).strip()
+                if line[0] == "*":
+                    words = line[1:].split()
+                    if words[0] in ["potential-index", "ipot"] and words[1] == "Z":
+                        return True
+                return False
+
+            elif line == "ATOMS":
+                line = next(generator).strip()
+                if line[0] == "*":
+                    words = line[1:].split()
+                    if words[:4] == ["x", "y", "z", "ipot"]:
+                        return True
+                return False
+
+            line = next(generator)
+
+        return False
+
+    def set_meta(self, dataset: "DatasetInstance", overwrite: bool = True, **kwd) -> None:
+        """
+        Extract metadata from TITLE
+        """
+        title_block = ""
+        headers = get_headers(dataset.file_name, sep=None, comment_designator="*")
+        for header in headers:
+            if header and header[0] == "TITLE":
+                title_block += " ".join(header[1:]) + "\n"
+
+        dataset.metadata.title_block = title_block
+
+    def set_peek(self, dataset: "DatasetInstance", **kwd) -> None:
+        if not dataset.dataset.purged:
+            dataset.peek = get_file_peek(dataset.file_name)
+            dataset.info = dataset.metadata.title_block
+
+        else:
+            dataset.peek = "file does not exist"
+            dataset.blurb = "file purged from disk"
diff --git a/lib/galaxy/datatypes/test/bad_atoms.inp b/lib/galaxy/datatypes/test/bad_atoms.inp
new file mode 100644
index 00000000000..b801beb1bd1
--- /dev/null
+++ b/lib/galaxy/datatypes/test/bad_atoms.inp
@@ -0,0 +1,44 @@
+TITLE comment: None given
+TITLE Source:  
+TITLE Structure Summary:  Fe2 S4
+TITLE Reduced formula:  FeS2
+TITLE space group: (Pnnm), space number:  (58)
+TITLE abc:  3.385200   4.447400   5.428700
+TITLE angles: 90.000000  90.000000  90.000000
+TITLE sites: 6
+* 1 Fe     0.000000     0.000000     0.000000
+* 2 Fe     0.500000     0.500000     0.500000
+* 3 S     0.000000     0.199900     0.378040
+* 4 S     0.000000     0.800100     0.621960
+* 5 S     0.500000     0.699900     0.121960
+* 6 S     0.500000     0.300100     0.878040
+
+ATOMS
+  *       a          b        c    ipot  Atom      Distance    Number
+**********-  ********-  ******-  ******  ******  **********  ********
+   0          0          0            0  Fe         0               0
+  -0.889035  -2.05227   -0            2  S          2.23656         4
+   0.889035   2.05227    0            2  S          2.23656        22
+  -1.33466    0.662084  -1.6926       2  S          2.2549          6
+  -1.33466    0.662084   1.6926       2  S          2.2549          8
+   1.33466   -0.662084  -1.6926       2  S          2.2549         11
+   1.33466   -0.662084   1.6926       2  S          2.2549         13
+   0          0         -3.3852       1  Fe         3.3852         18
+   0          0          3.3852       1  Fe         3.3852         19
+  -0.889035   3.37643    0            2  S          3.49152         9
+   0.889035  -3.37643   -0            2  S          3.49152        15
+  -3.11273   -0.662084  -1.6926       2  S          3.60449         1
+  -3.11273   -0.662084   1.6926       2  S          3.60449         3
+   3.11273    0.662084  -1.6926       2  S          3.60449        16
+   3.11273    0.662084   1.6926       2  S          3.60449        20
+  -2.2237    -2.71435   -1.6926       1  Fe         3.89582         2
+  -2.2237    -2.71435    1.6926       1  Fe         3.89582         5
+  -2.2237     2.71435   -1.6926       1  Fe         3.89582         7
+  -2.2237     2.71435    1.6926       1  Fe         3.89582        10
+   2.2237    -2.71435   -1.6926       1  Fe         3.89582        12
+   2.2237    -2.71435    1.6926       1  Fe         3.89582        14
+   2.2237     2.71435   -1.6926       1  Fe         3.89582        17
+   2.2237     2.71435    1.6926       1  Fe         3.89582        21
+END
+
+
diff --git a/lib/galaxy/datatypes/test/bad_potentials.inp b/lib/galaxy/datatypes/test/bad_potentials.inp
new file mode 100644
index 00000000000..e0c3a4f13f8
--- /dev/null
+++ b/lib/galaxy/datatypes/test/bad_potentials.inp
@@ -0,0 +1,24 @@
+TITLE comment: None given
+TITLE Source:  
+TITLE Structure Summary:  Fe2 S4
+TITLE Reduced formula:  FeS2
+TITLE space group: (Pnnm), space number:  (58)
+TITLE abc:  3.385200   4.447400   5.428700
+TITLE angles: 90.000000  90.000000  90.000000
+TITLE sites: 6
+* 1 Fe     0.000000     0.000000     0.000000
+* 2 Fe     0.500000     0.500000     0.500000
+* 3 S     0.000000     0.199900     0.378040
+* 4 S     0.000000     0.800100     0.621960
+* 5 S     0.500000     0.699900     0.121960
+* 6 S     0.500000     0.300100     0.878040
+
+POTENTIALS 
+  * pot    Z  tag      lmax1    lmax2    xnatph(stoichometry)    spinph
+******-  **-  ****-  ******-  ******-  **********************  ********
+      0   26  Fe          -1       -1                  0.0001         0
+      1   26  Fe          -1       -1                  2              0
+      2   16  S           -1       -1                  4              0
+END
+
+
diff --git a/lib/galaxy/datatypes/test/no_pymatgen.inp b/lib/galaxy/datatypes/test/no_pymatgen.inp
new file mode 100644
index 00000000000..d75015dee1c
--- /dev/null
+++ b/lib/galaxy/datatypes/test/no_pymatgen.inp
@@ -0,0 +1,51 @@
+TITLE comment: None given
+TITLE Source:  
+TITLE Structure Summary:  Fe2 S4
+TITLE Reduced formula:  FeS2
+TITLE space group: (Pnnm), space number:  (58)
+TITLE abc:  3.385200   4.447400   5.428700
+TITLE angles: 90.000000  90.000000  90.000000
+TITLE sites: 6
+* 1 Fe     0.000000     0.000000     0.000000
+* 2 Fe     0.500000     0.500000     0.500000
+* 3 S     0.000000     0.199900     0.378040
+* 4 S     0.000000     0.800100     0.621960
+* 5 S     0.500000     0.699900     0.121960
+* 6 S     0.500000     0.300100     0.878040
+
+POTENTIALS 
+  *ipot    Z  tag      lmax1    lmax2    xnatph(stoichometry)    spinph
+******-  **-  ****-  ******-  ******-  **********************  ********
+      0   26  Fe          -1       -1                  0.0001         0
+      1   26  Fe          -1       -1                  2              0
+      2   16  S           -1       -1                  4              0
+
+ATOMS
+  *       x          y        z    ipot  Atom      Distance    Number
+**********-  ********-  ******-  ******  ******  **********  ********
+   0          0          0            0  Fe         0               0
+  -0.889035  -2.05227   -0            2  S          2.23656         4
+   0.889035   2.05227    0            2  S          2.23656        22
+  -1.33466    0.662084  -1.6926       2  S          2.2549          6
+  -1.33466    0.662084   1.6926       2  S          2.2549          8
+   1.33466   -0.662084  -1.6926       2  S          2.2549         11
+   1.33466   -0.662084   1.6926       2  S          2.2549         13
+   0          0         -3.3852       1  Fe         3.3852         18
+   0          0          3.3852       1  Fe         3.3852         19
+  -0.889035   3.37643    0            2  S          3.49152         9
+   0.889035  -3.37643   -0            2  S          3.49152        15
+  -3.11273   -0.662084  -1.6926       2  S          3.60449         1
+  -3.11273   -0.662084   1.6926       2  S          3.60449         3
+   3.11273    0.662084  -1.6926       2  S          3.60449        16
+   3.11273    0.662084   1.6926       2  S          3.60449        20
+  -2.2237    -2.71435   -1.6926       1  Fe         3.89582         2
+  -2.2237    -2.71435    1.6926       1  Fe         3.89582         5
+  -2.2237     2.71435   -1.6926       1  Fe         3.89582         7
+  -2.2237     2.71435    1.6926       1  Fe         3.89582        10
+   2.2237    -2.71435   -1.6926       1  Fe         3.89582        12
+   2.2237    -2.71435    1.6926       1  Fe         3.89582        14
+   2.2237     2.71435   -1.6926       1  Fe         3.89582        17
+   2.2237     2.71435    1.6926       1  Fe         3.89582        21
+END
+
+
diff --git a/lib/galaxy/datatypes/test/test.inp b/lib/galaxy/datatypes/test/test.inp
new file mode 100644
index 00000000000..522743703c3
--- /dev/null
+++ b/lib/galaxy/datatypes/test/test.inp
@@ -0,0 +1,52 @@
+* This FEFF.inp file generated by pymatgen
+TITLE comment: None given
+TITLE Source:  
+TITLE Structure Summary:  Fe2 S4
+TITLE Reduced formula:  FeS2
+TITLE space group: (Pnnm), space number:  (58)
+TITLE abc:  3.385200   4.447400   5.428700
+TITLE angles: 90.000000  90.000000  90.000000
+TITLE sites: 6
+* 1 Fe     0.000000     0.000000     0.000000
+* 2 Fe     0.500000     0.500000     0.500000
+* 3 S     0.000000     0.199900     0.378040
+* 4 S     0.000000     0.800100     0.621960
+* 5 S     0.500000     0.699900     0.121960
+* 6 S     0.500000     0.300100     0.878040
+
+POTENTIALS 
+  *ipot    Z  tag      lmax1    lmax2    xnatph(stoichometry)    spinph
+******-  **-  ****-  ******-  ******-  **********************  ********
+      0   26  Fe          -1       -1                  0.0001         0
+      1   26  Fe          -1       -1                  2              0
+      2   16  S           -1       -1                  4              0
+
+ATOMS
+  *       x          y        z    ipot  Atom      Distance    Number
+**********-  ********-  ******-  ******  ******  **********  ********
+   0          0          0            0  Fe         0               0
+  -0.889035  -2.05227   -0            2  S          2.23656         4
+   0.889035   2.05227    0            2  S          2.23656        22
+  -1.33466    0.662084  -1.6926       2  S          2.2549          6
+  -1.33466    0.662084   1.6926       2  S          2.2549          8
+   1.33466   -0.662084  -1.6926       2  S          2.2549         11
+   1.33466   -0.662084   1.6926       2  S          2.2549         13
+   0          0         -3.3852       1  Fe         3.3852         18
+   0          0          3.3852       1  Fe         3.3852         19
+  -0.889035   3.37643    0            2  S          3.49152         9
+   0.889035  -3.37643   -0            2  S          3.49152        15
+  -3.11273   -0.662084  -1.6926       2  S          3.60449         1
+  -3.11273   -0.662084   1.6926       2  S          3.60449         3
+   3.11273    0.662084  -1.6926       2  S          3.60449        16
+   3.11273    0.662084   1.6926       2  S          3.60449        20
+  -2.2237    -2.71435   -1.6926       1  Fe         3.89582         2
+  -2.2237    -2.71435    1.6926       1  Fe         3.89582         5
+  -2.2237     2.71435   -1.6926       1  Fe         3.89582         7
+  -2.2237     2.71435    1.6926       1  Fe         3.89582        10
+   2.2237    -2.71435   -1.6926       1  Fe         3.89582        12
+   2.2237    -2.71435    1.6926       1  Fe         3.89582        14
+   2.2237     2.71435   -1.6926       1  Fe         3.89582        17
+   2.2237     2.71435    1.6926       1  Fe         3.89582        21
+END
+
+
diff --git a/lib/galaxy/datatypes/test/test.prj b/lib/galaxy/datatypes/test/test.prj
new file mode 100644
index 0000000000000000000000000000000000000000..55b0926c8718d2fd52ddd2a2bf5c50da946b48dd
GIT binary patch
literal 5372
zcmYkxbyO1$!0z!eT6%OyjvgueD5XYsjP4Fe0THA}3Id}W=?>|ZkP<efBm@*>ba(T5
z&U^1Y_k8|&&hur8$HP0l+pq&ZdHVVCd%bZ51e=+8f*0LG<{L;GF-C9<qkO(3ko9A%
z_t=wd44lN1;k~L$nXNHwZ|fF)(S&#^NK!m)6fp$uA09Vqog61KdCoP<SUgRjy`|mi
zt+Yn1whqv@A?<|?o*wm2ejx<{XFIohL6S#_F|WlgxJzpa>!0pBq|WLOTvp9DyZ4Z2
z$qrIa^A7Ka@6sWtgJ%DQTdS>=wYA$!q|4p4mKL{+)zj85G}(lB>v+s-oiF>u4Zl6t
zH&!-0dIUUNLr!OU8fLBxJ=D7pGJOa9Nn#{-E1B+DLA_nK6@MQOFV}p7f;%6#_OvUu
z9_Bp@AH6O{kvnUIBS+Hi#4&#zd<NDELmneT|I0l4_2Tiqo%ZQQ*t6iOduTGr*RQ)f
zmn3uL!J`jzek%2D;$S6?DwSXRd+rhQU*p4VK*NM*_t44-y<w6L;4Smx%+P`3lh?~N
zoYH~Ki#PYDV3M}?uUt*s3asvv62cT!20L@FGLZ*CA<HXPhV(}TFnoyPy5Z5@2F}*y
zW6R<P(dD%Cr;@z873{v=S5}*u4t9P;Ez5e`tA0i^3zrt{TuxsGa%0k|u106hf@~c|
z>W$u^&wJMQiG73e?QhQA+1GZS1UGHEGXfTF{cDarD1X<D<r!~Z7-WCi`x|E;<@P#o
zwuwF?NCMe{=IHQ$x%5{5)H?r|scYOi?`kwNplssC;V9Jj@i<H2R^-BbFh@$@8McCh
zA3St18U6IPgE>dy=?kVn_f4<UV|mf%;N=yMq`P^iIC|4U0nRq*x}>{T>k4d<?FBcB
zE1P2M{JHG^$@w@(h%qJNWgyY-Kuo_6yo!*yDmh1t+xMZIr-FrytHu;FN3bj_lr~&a
z$WI%gwhSg9y8=&e5}OrP3>X8&G~y|#QDMJk9Q_Nl#jG~Z0#+wxGV8X-<1>+HT~G0>
z%=Y0?Gke60V*nXGc_@RGz|6#sku8JvuN#-2QojomzqjYIQ^spjq3BN$2iLzqhAr=y
z?Q~AlFHkE5?YU??0d>)trjQVc0>H<E1+VLB&w}ysZ_@h9zfvm#T8V`2nFByIL+dXC
z&YSL}kK6-7vfI0nA5I=_473(J`yf(C(mUi$P*3n(*W#BC=e?_^0X~7Ih`(*ee^Ss~
zlGJq`Uy$7~qw#qs_g!E1Eib!G*8CP$TpF1QbO=!qi9-|eBO0A+$F~|Bubw4XUXD&k
z<?<)A|9yA}FeD(0x8RCDn{bp$9Ycn9bo_ByWz_appJVwwogdm~`87Y$YQQ@4PvLLN
zt)WYA^ON@Zl}lAVEyCHQ?qA58%a)t}#I4@O484PE1%?!m#Eyk~+Y}yRC4F!BayN5C
zz4;I{7;yHi1BUWgqY6yg$oD6SK|bc=r<MKaGRYd9pSgV1Ms`}`_2o~3!28Zk^hc{Q
za!V55Dk2#(8UYpRvxCg{7p1}ljH@po^R<>Gu(JZ6rxNDz)2|q-n>P1H{@uSkyB{7e
z3G9!O9APh$$I7tHS}!9EolOMdN_>wh+G?$g7I&_;<@yAO@I3KaPd9}xNGIfRW#&Ni
zGY+nz0`^$++<`^v#yrY9265#IPQXPg{00RbRvtv!m>LCpHApkAk5!mlS3*}p^&gXe
z6#p^!NAn*GiF^r5wS9TKQQlErPOg7={^9<Imy@@G>mQzfxc}i*_PeZZHp=ha-8XQ$
zpe7sC{omOj`F?J(24(sG)&FzJ`4Lz9=;Xt9Yk#oHi~sLrfNgz}#GR9<ME$?|f3Iqk
zu3uuc4=i8wS*$akFm1)shPS*wEPGO4RX(GtV-QT?8mH<!4NTQ3x{A{9;6aFiWzsp8
zW!I2koczf_kZI?1RwNNjUXhK{J~t(kD0n;Bg?XKgD>BtdSgU9pXF-xcOj~ny+%Mu*
zL*8SP?M>lzSOjgnm9pd*qA)|;F90{+Z4y#NVb7jnMVFR7e+ye=nA4V{P*AE0FPs;N
znshR7MqfvoRv1V9vL=?U9e}4>M8GH*a_a|tZL@KX$H<7Tb|FP_5q86p_oVMI^^718
zLfUWfk}8@OgdYea-NwK#X*0~QHd4ow25&la96`sF9MfV@7UpX+D-n`d?`Q*A@cd>r
zR&7&qqXno*GiWr7xhDO$GDP&J6tg{zG)DN8AFs7lHi3@alB2Me(1}^U`{o2zSgQ!j
znj=fN;&&V(fHj2!pdVrc*fm8HWy=nj|KT0?!-Kg6iBr&1@#0MGWPb*QePWEnaw~QV
z%U&cYlB7v9z?J`Cj@Nh`g$7bZLD>KVUM#7CF825US<+yPFpi(_dGZo&o%|Fcuhzck
zX^@F=>EM8}!gc%U0GTBvJ{kVYYxD?NK-UD$DZ3Syo|1!z;yaysTEyK;s~38XX9!iA
z^kF?sr&Pt#T)i7CHAsHvxXZmFIDRrL(M!kHUU37+jsP@~gA%oaMBhazvQ(;5f&zVl
zl*Ef$hegPWc7aS`*XM*yDstnU0(xL2LxwdYiF;c%Sz&4plKnlr)C!cLd^YtlC(mM#
zX+%DG_p1Svs6YjKB;hlXR6Q}!g;K(SLOGFqN}(|i4_pv&Kp-VKP*6arg0RhrVkN;P
zPQ>arLQ&##CAr<Djdw#55{v@`gzdaEfZyF|iDX%|7OdlB_l=De3pTJ2c%Ao#P~cjY
zx-4&sMCzC>0^2v@MSNLXZ1d~rUR&xEJx6;`_(+4pE~UI3NHw7$L$`~5R5v;syyKKr
z<t;aXc&|@N(HE}T^EGQ6mf4j|dHJU~130ATll_vqK0$}X{8rdHec^$&mD$H+EbVeJ
zp%VmqhZ#I#x93Cy78>z^I*{`NO5c-JhEEw+ezZeAL!0*n%BkQGM*w-#dU^~iJ=fZk
zUy)0->g<13jA#D_j1(b8MWoTtx?|gnbnwYWMfw&+j$ZL~1g`(!I(Luxt-O<3Y?Z+?
zN(Yn7z-IPsa>jeYB_nw9N2vn@>l^dJGM+c?XcV`6Zc7&?#Z@f>4L;V)#`6`H%TlG)
zpkuozqDR300*}1b1aBiBxl-M%#xdU?n=0kT7h0sCHuMPAE_3NSUg6fXiw{`3ijcge
z1t}U+(l7Gz*wNO|Zx^*5a8AtbiZ$11w8*aMqW(t?R#eu<B{@_Ck-M3-y4baV_d8M7
z2AUiqn}ZG(xv#FH(_AdV#&m$a$Xz^<)n^T=U9K<>1<0ZPHK>kF`wg}yMYUVVIi3EI
zygWTimCsh!RMpUC$5DTYlk?lz<H3r7hp?MsXiVtBLB6ZEvG2Zwv)<wMZ#uSy624Tu
zZjDcWrW0=3CbKrQ6Osd}_FBT+GhK;gE(*|aCcfmgH``5KjX9a_V{BOg_TdT9%Y%cY
zuGqk50-!@waCu3JvAMY*96OK@CM1E|*=x_aC7w_z+PDqu6}H1AD0OoNYd>4c+DEI7
zkkS(r>Nj$GH7gM@nwAxdG|?uM5>Skl`$9(!o4B|lZF4|B>;|#B!i@B%vFZMTw@T<q
z+GH;#?aUHbq%5!&eB20Vt3Pt$w;K}uAm_BQ4>xJ2GN8bu)n=MVjof%XGI$c(1YyXZ
zAo$IehU)6S8A#&fiufB>5}-iESM(yT3&1mr3kdmE&9&riUAj|jpmlRvx=6k-L+pn%
z>TQFb(3QW5UHOj?*k2`bp39YEOIm6JuuxD|2st?={${woA!mIeauolQh^BJa`h3mc
zU@8zQQ)dmee;=&F+VxYC$`>SH-LjlxO$WE~F|p3u`P;SSPDC8$e}f;<V|3_{3^sSo
ztElS!`m%F_we*$t3&zw+cN>>jSuqhNg6+7(e7ol&%&WzRG4UnGQ=jb7oRJ!bO}HE%
zSvig7sp_f%{PWW}=+JJ*n1}hJjEwUcuj;D`u#4JCWu|ef*%xOpMp0X~3~zKd01FPP
z@0m+#<Yuv6x;#4MOlwn1tC8eB_Z_|Frc%INm5PrGKb)ZZ<X+xyC&?D)(8fF{bb1e<
zYm%7P-HOt9uBdgD#o9C-a*Pr8>8RMel;@P*NjN2zO6^dMbGZ5$UZcwASPC{ES@<qv
z769p$c2^V^2VqE6Cm6!PrwkLIdDiAH#2#1$$yDSSJm+{1>Kft@KCwMv<&6BP=rCDE
zb01w}N=A?A8Wuaq&&k-}OyZF`>`5$Y@+odM62^U66yc+~vzh+&nyuU04TelMj{s1$
zdze3{zAv3(<}6cUdWDWwiH+dwH}g&i*9%H;jZP@>{a%J6%2Y-mR0Drfuu$;gR@5w1
zqc$pXf4l36>lu$PNGUIbh<p3Br6^pM<s;80{*$T_2l6wF9aIw6j#e>goVreU|B)h`
zh?EWkfBx-fT4Cp7pe_lfYltig38b=T`>pUEi^4Ixd>v69iLwOIPZ^&ha9_^r7n2*j
z{^;GI<Gj&Gqtt^ruO?P#pqUoR{~FB1=d1rb<!W7vX=kSC7(=fCkdOuy^~Z<^q}(X1
z-62wyn;Vb_`61m|zHWD^#TT0>2Q1)Uz1VfOnk{N{-3NkTX4RP*C6Z&B3`A7DYSYn1
zU*+%!C~#Fbxkm4WqMUhT<k^bs@uBS-QCuPh=50XP-C}FpmL#n!Hc4;>Gp^!r{C%u7
zWk~T+Zy-N~yTkTSolGx*DDD9w&4EPL48Ba@kqO!Ffi{lmCf>@jueo;eebY`Ug;&Se
zejjcsVewjawF_Tf&;OgUAY@eWGb(W`!h)*N^oKNXfF=yWoi{u+|9mk65pR*n=^Sc-
z;3rX~U_@x)2$U$ON~vLFH=hTMaWE29RYoVain^rem<#iK&cnPHc?^THscmIvj==;2
zy_<hX`ssC!v(9`fBZwpp1(ji6V6co2cB9r3u))`>=Wzsm6B64cq+eC3xf&HG7o8P6
zm)eY4g5+wW`M5%+i$z+Lm*~XoJz%7ZQPP5^TiUlcs>w9*#=R|4m9JpEXpHp=pLlr!
zWoe51&xMZuCZvKr#1txmflgcR5q&LhZ!uIf5-<7n+A;%xqQg5_fO86geD6B-7my>0
z>x6EAslDrhZ+Ij<I_JF4n06dEtR7^zb7?2yecZ=indje09V`Ur-B4KQPQ>R3tLxVJ
zsRl~^!8vCMQzf}KYA7L+|FM?`t*1f0z=(}eH^s`aX^~SjbJg4tuJ+3h3uAIum}+>-
zS-{QC(bBzyH>577Si-ABnWjbc(U(h7vy_4ida6~N$EF4SIl_-`7FJqxMq{U+zTTd|
z_F0_rV>u{J9Sg8M#{PzQU8G7b-e>mL8E{Pg1k~=NpgY|$9%&}<2L<bZ+QI?e9aXU-
zh~pAtTtz0E?I@+;3|+DZ`HCGCs__!Bv@@HtykQ-h#ZM6)aHkwB&Ms+L{dIf?g8wLF
zOUjfsy-)ipjs}D`nV(}^fpj{a4Vv|+Bhvk5OIG<$#b_a>n1}4MLV3l?<e}Era$&K*
zb#p8ZQ<$d;zwWYogJG<X;iFSYVwP7Mx`8_r6k&A3wsEA%?AV`*f3&Mxd~z_r>PLuB
z+nX@M{Ai%*d*?<`_TjtWX3|N`R0IS38;C+g5YmD$({pfBO_Y=75mBB14kbE@edSK(
z9qA879DrZ3&f&N+J75fne0H3q7(t{fxsFoWrYXu#kvFh7`QGk=321bj9Qvg}z8NH(
zQtd$!7+x2+=V3wYGL(Y=&5$Go!?9B~NHS$)<WSGOyWq2qw2u%q+qZhcG?{%fyxcPM
z`ip1qNt@(bbsmI!&!)S;Lt`bV%cH-tXxAQ$lH>UM?P!Qu_X)>jWR+E>IHI(nO>M;1
zo~e`xl%lE<Ov4kW>(USp-@_5h*0sB8vl%;Byu?A<X3@}Zw-1Ga$;RPQL5P(`%#W_{
zy-`c*^@Sgr^^1@$mLei#C3p`9iXUMqpzypq#W`mL$Lamhg4rlq-{kokET-jB@lEs#
z_{g^)=!I=Vk*ou>n(4X@2QA2)`N3wzy3kAzmIU8X=497=e*2&^9u6X`D-Wbm<CoFF
zuWlBrrQb1XY^R|^$&1svBtu2%?CGk97M{kC!8DYVm2?t+7L{c;rHkp&9Hz-1NvpgW
z5)7^P{h;!+KL>EG$mY`~-<r}6B>RjrEdz-<qgKWF5((j6bUE3W!a7oy_12#Bw)jl@
z7a2SeMV^#_tHR;U4CrXxRNC|<9+s&!S`kQT37RWDKx1WfaC~sqtY=Pm%n>)qO<`u6
zbTdZMZvm2S(+s;K{fW!5Gc(W3x41#4;nB3sSiwFaWz)~-@SY(rI!X61I#Qfa(ID7M
z8LL)I!4otI9@vN$Olx_kK9EB5a6rngJgjQim)tt);<3zEuW_K10c|?kwEMe+eh|bD
zR@M>DP8q0Wj0v`*--Z@WCA7>a(Un@^OzzfL$5MH+c{N3a9#WZHnhNL6iObU0wt?Xf
z)g$GM(z1S*bMl@TCUuT=zJkFPt?yoQ%md$0P7h3{3m7^sJ)5&@XkI@wZGCOpBPzAf
zAiwmj&ML~KoyZV?45FEo!9z?R{6V1g&W10EH8OO|b~ss^hs2Q#@m0tTx{y?SvlxAm
z`kiFHy4uJxTN^q)uSv%#OM=Us7s$ZYdUyNd^~ul9kGQBjgE3w256y;15O>$SL80rL
zdCF-^pP{prQ_mOgaxk4Y=u4k3*XFd@QVQ37o3AE%eq#8m8p5%sGKAasy)#$`GW5;y
zXQ2sQslazClLNS*rr{P6x=5t&w0AJS^4d6i>DxiIeBh8tI|HQ>hpe54kbT(aFf5Ur
zDm!zhGL&s4dyUqVn#E^d2lf^vwz@0x_99HWKH{cdWWzb3x4te}47A>&QYP-)tNgG$
zQN#*3(qP_E1sM(d`;&nvbuP>~tEC+Bj{ZBL)LgElXi#$cYr!VtrM4p0cFKNtXcg0J
zhW!tfnmAVjv)b#gLed-CM$69V365$?M#7f4;8@icaq&O*`D{L9-!$BZlHXrlXwMrN
zh2CFW2THnKW)(j@%7k`wonBw<R_s|NnjJqjNgQl@UKYy0{UmxB+j1v@piiDFH)j{v
VJu=S^rL7(WmaZmMLc{@p{{dPeLq7lj

literal 0
HcmV?d00001


From 9a886cfc2460c1f223185dcd20f635288192fa3e Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Tue, 6 Jun 2023 12:54:55 +0000
Subject: [PATCH 2/9] Handle StopIteration in FEFFInput sniffer

---
 lib/galaxy/datatypes/larch.py                 | 55 ++++++++++++-------
 .../test/{no_pymatgen.inp => feff_atoms.inp}  |  7 ---
 .../{bad_atoms.inp => feff_bad_atoms.inp}     |  0
 ...potentials.inp => feff_bad_potentials.inp} |  0
 lib/galaxy/datatypes/test/feff_potentials.inp | 24 ++++++++
 .../test/{test.inp => feff_pymatgen.inp}      |  0
 6 files changed, 58 insertions(+), 28 deletions(-)
 rename lib/galaxy/datatypes/test/{no_pymatgen.inp => feff_atoms.inp} (85%)
 rename lib/galaxy/datatypes/test/{bad_atoms.inp => feff_bad_atoms.inp} (100%)
 rename lib/galaxy/datatypes/test/{bad_potentials.inp => feff_bad_potentials.inp} (100%)
 create mode 100644 lib/galaxy/datatypes/test/feff_potentials.inp
 rename lib/galaxy/datatypes/test/{test.inp => feff_pymatgen.inp} (100%)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index 80f7fb358c0..43a738e6ba9 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -70,6 +70,10 @@ class AthenaProject(Text):
         >>> fname = get_test_fname('test.prj')
         >>> AthenaProject().sniff(fname)
         True
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('Si.cif')
+        >>> FEFFInput().sniff(fname)
+        False
         """
 
         return file_prefix.startswith("# Athena project file")
@@ -135,19 +139,23 @@ class FEFFInput(Text):
         Try to guess if the file is an FEFF input file.
 
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('test.inp')
+        >>> fname = get_test_fname('feff_pymatgen.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('no_pymatgen.inp')
+        >>> fname = get_test_fname('feff_atoms.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('bad_atoms.inp')
+        >>> fname = get_test_fname('feff_potentials.inp')
+        >>> FEFFInput().sniff(fname)
+        True
+        >>> from galaxy.datatypes.sniff import get_test_fname
+        >>> fname = get_test_fname('feff_bad_atoms.inp')
         >>> FEFFInput().sniff(fname)
         False
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('bad_potential.inp')
+        >>> fname = get_test_fname('feff_bad_potentials.inp')
         >>> FEFFInput().sniff(fname)
         False
         >>> from galaxy.datatypes.sniff import get_test_fname
@@ -160,25 +168,30 @@ class FEFFInput(Text):
             return True
 
         generator = file_prefix.line_iterator()
-        line = next(generator)
-        while line is not None:
-            if line == "POTENTIALS":
-                line = next(generator).strip()
-                if line[0] == "*":
-                    words = line[1:].split()
-                    if words[0] in ["potential-index", "ipot"] and words[1] == "Z":
-                        return True
-                return False
+        try:
+            line = next(generator).strip()
+            while line is not None:
+                if line == "POTENTIALS":
+                    line = next(generator).strip()
+                    if line[0] == "*":
+                        words = line[1:].split()
+                        if (words[0] in ["potential-index", "ipot"]) and (words[1] == "Z"):
+                            return True
+                    return False
 
-            elif line == "ATOMS":
-                line = next(generator).strip()
-                if line[0] == "*":
-                    words = line[1:].split()
-                    if words[:4] == ["x", "y", "z", "ipot"]:
-                        return True
-                return False
+                elif line == "ATOMS":
+                    line = next(generator).strip()
+                    if line[0] == "*":
+                        words = line[1:].split()
+                        if words[:4] == ["x", "y", "z", "ipot"]:
+                            return True
+                    return False
 
-            line = next(generator)
+                else:
+                    line = next(generator).strip()
+
+        except StopIteration:
+            return False
 
         return False
 
diff --git a/lib/galaxy/datatypes/test/no_pymatgen.inp b/lib/galaxy/datatypes/test/feff_atoms.inp
similarity index 85%
rename from lib/galaxy/datatypes/test/no_pymatgen.inp
rename to lib/galaxy/datatypes/test/feff_atoms.inp
index d75015dee1c..6deb2fb4635 100644
--- a/lib/galaxy/datatypes/test/no_pymatgen.inp
+++ b/lib/galaxy/datatypes/test/feff_atoms.inp
@@ -13,13 +13,6 @@ TITLE sites: 6
 * 5 S     0.500000     0.699900     0.121960
 * 6 S     0.500000     0.300100     0.878040
 
-POTENTIALS 
-  *ipot    Z  tag      lmax1    lmax2    xnatph(stoichometry)    spinph
-******-  **-  ****-  ******-  ******-  **********************  ********
-      0   26  Fe          -1       -1                  0.0001         0
-      1   26  Fe          -1       -1                  2              0
-      2   16  S           -1       -1                  4              0
-
 ATOMS
   *       x          y        z    ipot  Atom      Distance    Number
 **********-  ********-  ******-  ******  ******  **********  ********
diff --git a/lib/galaxy/datatypes/test/bad_atoms.inp b/lib/galaxy/datatypes/test/feff_bad_atoms.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/bad_atoms.inp
rename to lib/galaxy/datatypes/test/feff_bad_atoms.inp
diff --git a/lib/galaxy/datatypes/test/bad_potentials.inp b/lib/galaxy/datatypes/test/feff_bad_potentials.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/bad_potentials.inp
rename to lib/galaxy/datatypes/test/feff_bad_potentials.inp
diff --git a/lib/galaxy/datatypes/test/feff_potentials.inp b/lib/galaxy/datatypes/test/feff_potentials.inp
new file mode 100644
index 00000000000..cc96ad4cd11
--- /dev/null
+++ b/lib/galaxy/datatypes/test/feff_potentials.inp
@@ -0,0 +1,24 @@
+TITLE comment: None given
+TITLE Source:  
+TITLE Structure Summary:  Fe2 S4
+TITLE Reduced formula:  FeS2
+TITLE space group: (Pnnm), space number:  (58)
+TITLE abc:  3.385200   4.447400   5.428700
+TITLE angles: 90.000000  90.000000  90.000000
+TITLE sites: 6
+* 1 Fe     0.000000     0.000000     0.000000
+* 2 Fe     0.500000     0.500000     0.500000
+* 3 S     0.000000     0.199900     0.378040
+* 4 S     0.000000     0.800100     0.621960
+* 5 S     0.500000     0.699900     0.121960
+* 6 S     0.500000     0.300100     0.878040
+
+POTENTIALS 
+  *ipot    Z  tag      lmax1    lmax2    xnatph(stoichometry)    spinph
+******-  **-  ****-  ******-  ******-  **********************  ********
+      0   26  Fe          -1       -1                  0.0001         0
+      1   26  Fe          -1       -1                  2              0
+      2   16  S           -1       -1                  4              0
+END
+
+
diff --git a/lib/galaxy/datatypes/test/test.inp b/lib/galaxy/datatypes/test/feff_pymatgen.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/test.inp
rename to lib/galaxy/datatypes/test/feff_pymatgen.inp

From 2efae4c392fb98b2ea512fcc303464e34940fe8a Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Tue, 6 Jun 2023 15:26:36 +0000
Subject: [PATCH 3/9] Rename test files to avoid guessing feff datatype

---
 lib/galaxy/datatypes/larch.py                          | 10 +++++-----
 .../datatypes/test/{feff_atoms.inp => inp_atoms.inp}   |  0
 .../test/{feff_bad_atoms.inp => inp_bad_atoms.inp}     |  0
 ...{feff_bad_potentials.inp => inp_bad_potentials.inp} |  0
 .../test/{feff_potentials.inp => inp_potentials.inp}   |  0
 .../test/{feff_pymatgen.inp => inp_pymatgen.inp}       |  0
 6 files changed, 5 insertions(+), 5 deletions(-)
 rename lib/galaxy/datatypes/test/{feff_atoms.inp => inp_atoms.inp} (100%)
 rename lib/galaxy/datatypes/test/{feff_bad_atoms.inp => inp_bad_atoms.inp} (100%)
 rename lib/galaxy/datatypes/test/{feff_bad_potentials.inp => inp_bad_potentials.inp} (100%)
 rename lib/galaxy/datatypes/test/{feff_potentials.inp => inp_potentials.inp} (100%)
 rename lib/galaxy/datatypes/test/{feff_pymatgen.inp => inp_pymatgen.inp} (100%)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index 43a738e6ba9..6494776e8cc 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -139,23 +139,23 @@ class FEFFInput(Text):
         Try to guess if the file is an FEFF input file.
 
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('feff_pymatgen.inp')
+        >>> fname = get_test_fname('inp_pymatgen.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('feff_atoms.inp')
+        >>> fname = get_test_fname('inp_atoms.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('feff_potentials.inp')
+        >>> fname = get_test_fname('inp_potentials.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('feff_bad_atoms.inp')
+        >>> fname = get_test_fname('inp_bad_atoms.inp')
         >>> FEFFInput().sniff(fname)
         False
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('feff_bad_potentials.inp')
+        >>> fname = get_test_fname('inp_bad_potentials.inp')
         >>> FEFFInput().sniff(fname)
         False
         >>> from galaxy.datatypes.sniff import get_test_fname
diff --git a/lib/galaxy/datatypes/test/feff_atoms.inp b/lib/galaxy/datatypes/test/inp_atoms.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/feff_atoms.inp
rename to lib/galaxy/datatypes/test/inp_atoms.inp
diff --git a/lib/galaxy/datatypes/test/feff_bad_atoms.inp b/lib/galaxy/datatypes/test/inp_bad_atoms.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/feff_bad_atoms.inp
rename to lib/galaxy/datatypes/test/inp_bad_atoms.inp
diff --git a/lib/galaxy/datatypes/test/feff_bad_potentials.inp b/lib/galaxy/datatypes/test/inp_bad_potentials.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/feff_bad_potentials.inp
rename to lib/galaxy/datatypes/test/inp_bad_potentials.inp
diff --git a/lib/galaxy/datatypes/test/feff_potentials.inp b/lib/galaxy/datatypes/test/inp_potentials.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/feff_potentials.inp
rename to lib/galaxy/datatypes/test/inp_potentials.inp
diff --git a/lib/galaxy/datatypes/test/feff_pymatgen.inp b/lib/galaxy/datatypes/test/inp_pymatgen.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/feff_pymatgen.inp
rename to lib/galaxy/datatypes/test/inp_pymatgen.inp

From 44bf8f5007ac7ba0f9764822e6186a483477ef7a Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Wed, 7 Jun 2023 08:39:12 +0000
Subject: [PATCH 4/9] Rename bad test files to avoid guessing inp datatype

---
 lib/galaxy/datatypes/test/{inp_atoms.inp => larch_atoms.inp}      | 0
 .../datatypes/test/{inp_bad_atoms.inp => larch_bad_atoms.txt}     | 0
 .../test/{inp_bad_potentials.inp => larch_bad_potentials.txt}     | 0
 .../datatypes/test/{inp_potentials.inp => larch_potentials.inp}   | 0
 .../datatypes/test/{inp_pymatgen.inp => larch_pymatgen.inp}       | 0
 5 files changed, 0 insertions(+), 0 deletions(-)
 rename lib/galaxy/datatypes/test/{inp_atoms.inp => larch_atoms.inp} (100%)
 rename lib/galaxy/datatypes/test/{inp_bad_atoms.inp => larch_bad_atoms.txt} (100%)
 rename lib/galaxy/datatypes/test/{inp_bad_potentials.inp => larch_bad_potentials.txt} (100%)
 rename lib/galaxy/datatypes/test/{inp_potentials.inp => larch_potentials.inp} (100%)
 rename lib/galaxy/datatypes/test/{inp_pymatgen.inp => larch_pymatgen.inp} (100%)

diff --git a/lib/galaxy/datatypes/test/inp_atoms.inp b/lib/galaxy/datatypes/test/larch_atoms.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/inp_atoms.inp
rename to lib/galaxy/datatypes/test/larch_atoms.inp
diff --git a/lib/galaxy/datatypes/test/inp_bad_atoms.inp b/lib/galaxy/datatypes/test/larch_bad_atoms.txt
similarity index 100%
rename from lib/galaxy/datatypes/test/inp_bad_atoms.inp
rename to lib/galaxy/datatypes/test/larch_bad_atoms.txt
diff --git a/lib/galaxy/datatypes/test/inp_bad_potentials.inp b/lib/galaxy/datatypes/test/larch_bad_potentials.txt
similarity index 100%
rename from lib/galaxy/datatypes/test/inp_bad_potentials.inp
rename to lib/galaxy/datatypes/test/larch_bad_potentials.txt
diff --git a/lib/galaxy/datatypes/test/inp_potentials.inp b/lib/galaxy/datatypes/test/larch_potentials.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/inp_potentials.inp
rename to lib/galaxy/datatypes/test/larch_potentials.inp
diff --git a/lib/galaxy/datatypes/test/inp_pymatgen.inp b/lib/galaxy/datatypes/test/larch_pymatgen.inp
similarity index 100%
rename from lib/galaxy/datatypes/test/inp_pymatgen.inp
rename to lib/galaxy/datatypes/test/larch_pymatgen.inp

From fb64b7d989601a0a1446922d45e1cb33e5117118 Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Wed, 7 Jun 2023 09:38:19 +0000
Subject: [PATCH 5/9] Edit test filenames in larch.py

---
 lib/galaxy/datatypes/larch.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index 6494776e8cc..28715e8695c 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -139,23 +139,23 @@ class FEFFInput(Text):
         Try to guess if the file is an FEFF input file.
 
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('inp_pymatgen.inp')
+        >>> fname = get_test_fname('larch_pymatgen.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('inp_atoms.inp')
+        >>> fname = get_test_fname('larch_atoms.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('inp_potentials.inp')
+        >>> fname = get_test_fname('larch_potentials.inp')
         >>> FEFFInput().sniff(fname)
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('inp_bad_atoms.inp')
+        >>> fname = get_test_fname('larch_bad_atoms.txt')
         >>> FEFFInput().sniff(fname)
         False
         >>> from galaxy.datatypes.sniff import get_test_fname
-        >>> fname = get_test_fname('inp_bad_potentials.inp')
+        >>> fname = get_test_fname('larch_bad_potentials.txt')
         >>> FEFFInput().sniff(fname)
         False
         >>> from galaxy.datatypes.sniff import get_test_fname

From 1b24e52c6e21117116922e57a7a3d1f19934f070 Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Fri, 4 Aug 2023 17:09:15 +0000
Subject: [PATCH 6/9] Allow unset metadata for athena project

---
 lib/galaxy/datatypes/larch.py | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index 28715e8695c..75118f0f99d 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -82,24 +82,26 @@ class AthenaProject(Text):
         """
         Extract metadata from @args
         """
+        def extract_arg(args: "list[str]", arg_name: str):
+            try:
+                index = args.index(f"'{arg_name}'")
+                setattr(dataset.metadata, arg_name, args[index + 1].replace("'", ""))
+            except ValueError:
+                return
+
         headers = get_headers(dataset.file_name, sep=" = ", count=3, comment_designator="#")
+        args = []
         for header in headers:
             if header[0] == "@args":
                 args = header[1][1:-2].split(",")
                 break
 
-        index = args.index("'atsym'")
-        dataset.metadata.atsym = args[index + 1][1:-1]
-        index = args.index("'bkg_e0'")
-        dataset.metadata.bkg_e0 = args[index + 1][1:-1]
-        index = args.index("'edge'")
-        dataset.metadata.edge = args[index + 1][1:-1]
-        index = args.index("'npts'")
-        dataset.metadata.npts = args[index + 1][1:-1]
-        index = args.index("'xmax'")
-        dataset.metadata.xmax = args[index + 1][1:-1]
-        index = args.index("'xmin'")
-        dataset.metadata.xmin = args[index + 1][1:-1]
+        extract_arg(args, "atsym")
+        extract_arg(args, "bkg_e0")
+        extract_arg(args, "edge")
+        extract_arg(args, "npts")
+        extract_arg(args, "xmax")
+        extract_arg(args, "xmin")
 
     def set_peek(self, dataset: "DatasetInstance", **kwd) -> None:
         if not dataset.dataset.purged:

From d8a80eff49148405a2c34dced448285b559fe921 Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Fri, 24 Nov 2023 11:39:23 +0000
Subject: [PATCH 7/9] Fix typo in AthenaProject sniffer

---
 lib/galaxy/datatypes/larch.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index 75118f0f99d..25248be09c6 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -72,7 +72,7 @@ class AthenaProject(Text):
         True
         >>> from galaxy.datatypes.sniff import get_test_fname
         >>> fname = get_test_fname('Si.cif')
-        >>> FEFFInput().sniff(fname)
+        >>> AthenaProject().sniff(fname)
         False
         """
 

From 36616a822794e9cb632b17ac4e10412db585dce8 Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Fri, 24 Nov 2023 12:02:01 +0000
Subject: [PATCH 8/9] Use List from typing for typehint in larch.py

---
 lib/galaxy/datatypes/larch.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index 25248be09c6..c91231af9c2 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING
+from typing import List, TYPE_CHECKING
 
 from galaxy.datatypes.data import (
     get_file_peek,
@@ -82,7 +82,8 @@ class AthenaProject(Text):
         """
         Extract metadata from @args
         """
-        def extract_arg(args: "list[str]", arg_name: str):
+
+        def extract_arg(args: List[str], arg_name: str):
             try:
                 index = args.index(f"'{arg_name}'")
                 setattr(dataset.metadata, arg_name, args[index + 1].replace("'", ""))

From cfab9e8169c44f5eae11c0ffedaaf83524273c7b Mon Sep 17 00:00:00 2001
From: Patrick Austin <patrick.austin@stfc.ac.uk>
Date: Fri, 24 Nov 2023 13:23:15 +0000
Subject: [PATCH 9/9] DatasetInstance -> DatasetProtocol in larch.py typehints

---
 lib/galaxy/datatypes/larch.py | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/lib/galaxy/datatypes/larch.py b/lib/galaxy/datatypes/larch.py
index c91231af9c2..212f41af1a0 100644
--- a/lib/galaxy/datatypes/larch.py
+++ b/lib/galaxy/datatypes/larch.py
@@ -1,19 +1,17 @@
-from typing import List, TYPE_CHECKING
+from typing import List
 
 from galaxy.datatypes.data import (
     get_file_peek,
     Text,
 )
 from galaxy.datatypes.metadata import MetadataElement
+from galaxy.datatypes.protocols import DatasetProtocol
 from galaxy.datatypes.sniff import (
     build_sniff_from_prefix,
     FilePrefix,
     get_headers,
 )
 
-if TYPE_CHECKING:
-    from galaxy.model import DatasetInstance
-
 
 @build_sniff_from_prefix
 class AthenaProject(Text):
@@ -78,7 +76,7 @@ class AthenaProject(Text):
 
         return file_prefix.startswith("# Athena project file")
 
-    def set_meta(self, dataset: "DatasetInstance", overwrite: bool = True, **kwd) -> None:
+    def set_meta(self, dataset: DatasetProtocol, *, overwrite: bool = True, **kwd) -> None:
         """
         Extract metadata from @args
         """
@@ -104,7 +102,7 @@ class AthenaProject(Text):
         extract_arg(args, "xmax")
         extract_arg(args, "xmin")
 
-    def set_peek(self, dataset: "DatasetInstance", **kwd) -> None:
+    def set_peek(self, dataset: DatasetProtocol, **kwd) -> None:
         if not dataset.dataset.purged:
             dataset.peek = get_file_peek(dataset.file_name)
             dataset.info = (
@@ -198,7 +196,7 @@ class FEFFInput(Text):
 
         return False
 
-    def set_meta(self, dataset: "DatasetInstance", overwrite: bool = True, **kwd) -> None:
+    def set_meta(self, dataset: DatasetProtocol, overwrite: bool = True, **kwd) -> None:
         """
         Extract metadata from TITLE
         """
@@ -210,7 +208,7 @@ class FEFFInput(Text):
 
         dataset.metadata.title_block = title_block
 
-    def set_peek(self, dataset: "DatasetInstance", **kwd) -> None:
+    def set_peek(self, dataset: DatasetProtocol, **kwd) -> None:
         if not dataset.dataset.purged:
             dataset.peek = get_file_peek(dataset.file_name)
             dataset.info = dataset.metadata.title_block