Make sure repository files displayed during installation are restricted to a max file size and contents are viewable in a browser.

lib/galaxy/util/shed_util.py

@@ -1,4 +1,4 @@
-import os, tempfile, shutil, subprocess, logging
+import os, tempfile, shutil, subprocess, logging, string
 from datetime import date, datetime, timedelta
 from time import strftime
 from galaxy import util
@@ -13,6 +13,14 @@ from elementtree.ElementTree import Element, SubElement
 
 log = logging.getLogger( __name__ )
 
+# Characters that must be html escaped
+MAPPED_CHARS = { '>' :'>', 
+                 '<' :'&lt;',
+                 '"' : '&quot;',
+                 '&' : '&amp;',
+                 '\'' : '&apos;' }
+VALID_CHARS = set( string.letters + string.digits + "'\"-=_.()/+*^,:?!#[]%\\$@;{}" )
+
 def add_to_shed_tool_config( app, shed_tool_conf_dict, elem_list ):
     # A tool shed repository is being installed so change the shed_tool_conf file.  Parse the config file to generate the entire list
     # of config_elems instead of using the in-memory list since it will be a subset of the entire list if one or more repositories have
@@ -1096,6 +1104,34 @@ def remove_from_tool_panel( trans, repository, shed_tool_conf, uninstall ):
     if uninstall:
         # Write the current in-memory version of the integrated_tool_panel.xml file to disk.
         trans.app.toolbox.write_integrated_tool_panel_config_file()
+def to_html_escaped( text ):
+    """Translates the characters in text to html values"""
+    translated = []
+    for c in text:
+        if c in [ '\r\n', '\n', ' ', '\t' ] or c in VALID_CHARS:
+            translated.append( c )
+        elif c in MAPPED_CHARS:
+            translated.append( MAPPED_CHARS[ c ] )
+        else:
+            translated.append( 'X' )
+    return ''.join( translated )
+def to_html_str( text ):
+    """Translates the characters in text to sn html string"""
+    translated = []
+    for c in text:
+        if c in VALID_CHARS:
+            translated.append( c )
+        elif c in MAPPED_CHARS:
+            translated.append( MAPPED_CHARS[ c ] )
+        elif c == ' ':
+            translated.append( '&nbsp;' )
+        elif c == '\t':
+            translated.append( '&nbsp;&nbsp;&nbsp;&nbsp;' )
+        elif c == '\n':
+            translated.append( '<br/>' )
+        elif c not in [ '\r' ]:
+            translated.append( 'X' )
+    return ''.join( translated )
 def update_repository( current_working_dir, repo_files_dir, changeset_revision ):
     # Update the cloned repository to changeset_revision.  It is imperative that the 
     # installed repository is updated to the desired changeset_revision before metadata

lib/galaxy/web/controllers/admin_toolshed.py

@@ -6,6 +6,8 @@ from galaxy import tools
 
 log = logging.getLogger( __name__ )
 
+MAX_CONTENT_SIZE = 32768
+
 class RepositoryListGrid( grids.Grid ):
     class NameColumn( grids.TextColumn ):
         def get_value( self, trans, grid, tool_shed_repository ):
@@ -377,8 +379,15 @@ class AdminToolshed( AdminGalaxy ):
                 owner = get_repository_owner( clean_repository_clone_url( repository_clone_url ) )
                 url = '%s/repository/get_readme?name=%s&owner=%s&changeset_revision=%s&webapp=galaxy' % ( tool_shed_url, name, owner, changeset_revision )
                 response = urllib2.urlopen( url )
-                readme_text = response.read()
+                raw_text = response.read()
                 response.close()
+                readme_text = ''
+                for i, line in enumerate( raw_text ):
+                    readme_text = '%s%s' % ( readme_text, to_html_str( line ) )
+                    if len( readme_text ) > MAX_CONTENT_SIZE:
+                        large_str = '\nFile contents truncated because file size is larger than maximum viewing size of %s\n' % util.nice_size( MAX_CONTENT_SIZE )
+                        readme_text = '%s%s' % ( readme_text, to_html_str( large_str ) )
+                        break
             else:
                 readme_text = '' 
         else:

lib/galaxy/webapps/community/controllers/common.py

@@ -6,7 +6,7 @@ from galaxy.tools import *
 from galaxy.util.json import from_json_string, to_json_string
 from galaxy.util.hash_util import *
 from galaxy.util.shed_util import copy_sample_loc_file, generate_datatypes_metadata, generate_tool_metadata, generate_workflow_metadata
-from galaxy.util.shed_util import handle_sample_tool_data_table_conf_file
+from galaxy.util.shed_util import handle_sample_tool_data_table_conf_file, to_html_escaped, to_html_str
 from galaxy.web.base.controller import *
 from galaxy.webapps.community import model
 from galaxy.model.orm import *

lib/galaxy/webapps/community/controllers/repository.py

@@ -15,14 +15,7 @@ from mercurial import hg, ui, patch, commands
 
 log = logging.getLogger( __name__ )
 
-# Characters that must be html escaped
-MAPPED_CHARS = { '>' :'>', 
-                 '<' :'&lt;',
-                 '"' : '&quot;',
-                 '&' : '&amp;',
-                 '\'' : '&apos;' }
 MAX_CONTENT_SIZE = 32768
-VALID_CHARS = set( string.letters + string.digits + "'\"-=_.()/+*^,:?!#[]%\\$@;{}" )
 VALID_REPOSITORYNAME_RE = re.compile( "^[a-z0-9\_]+$" )
     
 class CategoryListGrid( grids.Grid ):
@@ -785,7 +778,7 @@ class RepositoryController( BaseUIController, ItemRatings ):
         repo = hg.repository( get_configured_ui(), repo_dir )
         for root, dirs, files in os.walk( repo_dir ):
             for name in files:
-                if name.lower() in [ 'readme', 'read_me' ]:
+                if name.lower() in [ 'readme', 'readme.txt', 'read_me', 'read_me.txt' ]:
                     f = open( os.path.join( root, name ), 'r' )
                     text = f.read()
                     f.close()
@@ -1600,7 +1593,7 @@ class RepositoryController( BaseUIController, ItemRatings ):
         anchors = modified + added + removed + deleted + unknown + ignored + clean
         diffs = []
         for diff in patch.diff( repo, node1=ctx_parent.node(), node2=ctx.node() ):
-            diffs.append( self.to_html_escaped( diff ) )
+            diffs.append( to_html_escaped( diff ) )
         is_malicious = change_set_is_malicious( trans, id, repository.tip )
         return trans.fill_template( '/webapps/community/repository/view_changeset.mako', 
                                     repository=repository,
@@ -1938,20 +1931,20 @@ class RepositoryController( BaseUIController, ItemRatings ):
         trans.response.headers['Pragma'] = 'no-cache'
         trans.response.headers['Expires'] = '0'
         if is_gzip( file_path ):
-            to_html = self.to_html_str( '\ngzip compressed file\n' )
+            to_html = to_html_str( '\ngzip compressed file\n' )
         elif is_bz2( file_path ):
-            to_html = self.to_html_str( '\nbz2 compressed file\n' )
+            to_html = to_html_str( '\nbz2 compressed file\n' )
         elif check_zip( file_path ):
-            to_html = self.to_html_str( '\nzip compressed file\n' )
+            to_html = to_html_str( '\nzip compressed file\n' )
         elif check_binary( file_path ):
-            to_html = self.to_html_str( '\nBinary file\n' )
+            to_html = to_html_str( '\nBinary file\n' )
         else:
             to_html = ''
             for i, line in enumerate( open( file_path ) ):
-                to_html = '%s%s' % ( to_html, self.to_html_str( line ) )
+                to_html = '%s%s' % ( to_html, to_html_str( line ) )
                 if len( to_html ) > MAX_CONTENT_SIZE:
                     large_str = '\nFile contents truncated because file size is larger than maximum viewing size of %s\n' % util.nice_size( MAX_CONTENT_SIZE )
-                    to_html = '%s%s' % ( to_html, self.to_html_str( large_str ) )
+                    to_html = '%s%s' % ( to_html, to_html_str( large_str ) )
                     break
         return to_html
     @web.expose
@@ -1960,34 +1953,6 @@ class RepositoryController( BaseUIController, ItemRatings ):
         message = util.restore_text( params.get( 'message', ''  ) )
         status = params.get( 'status', 'done' )
         return trans.fill_template( '/webapps/community/repository/help.mako', message=message, status=status, **kwd )
-    def to_html_escaped( self, text ):
-        """Translates the characters in text to html values"""
-        translated = []
-        for c in text:
-            if c in [ '\r\n', '\n', ' ', '\t' ] or c in VALID_CHARS:
-                translated.append( c )
-            elif c in MAPPED_CHARS:
-                translated.append( MAPPED_CHARS[ c ] )
-            else:
-                translated.append( 'X' )
-        return ''.join( translated )
-    def to_html_str( self, text ):
-        """Translates the characters in text to sn html string"""
-        translated = []
-        for c in text:
-            if c in VALID_CHARS:
-                translated.append( c )
-            elif c in MAPPED_CHARS:
-                translated.append( MAPPED_CHARS[ c ] )
-            elif c == ' ':
-                translated.append( '&nbsp;' )
-            elif c == '\t':
-                translated.append( '&nbsp;&nbsp;&nbsp;&nbsp;' )
-            elif c == '\n':
-                translated.append( '<br/>' )
-            elif c not in [ '\r' ]:
-                translated.append( 'X' )
-        return ''.join( translated )
     def __build_allow_push_select_field( self, trans, current_push_list, selected_value='none' ):
         options = []
         for user in trans.sa_session.query( trans.model.User ):