Various fixes & enhancements

8 files changed, 136 insertions, 63 deletions

diff --git a/src/packages/changelogs.py b/src/packages/changelogs.py
index 53335c760b..606572429e 100644
--- a/src/packages/changelogs.py
+++ b/src/packages/changelogs.py
@@ -4,49 +4,49 @@ import mstring
 import re
 
 
-BUG_REGEX = re.compile(r'#[0-9]+|bug [0-9]+',re.I)
+BUG_REGEX = re.compile(r'\b#[0-9]+|\bbug [0-9]+',re.I)
 BUG_URL = 'http://bugs.gentoo.org/show_bug.cgi?id='
 
 def bugs_to_html(s):
-	index = 1
-	while 1:
-		match = BUG_REGEX.search(s,index)
-		if not match:
-			break
-		start = match.start()
-		end = match.end()
-		substring = s[start:end]
-		if substring[0] == '#':  # this of the form "#1234"
-			bugid = substring[1:]
-		else: # this is of the form "bug 1234"
-			bugid = substring[4:]
-		url = '<a href="%s%s">%s</a>' % (BUG_URL,bugid,substring)
-		(s,index) = mstring.replace_sub(s,url,start,end-1)
-	return s
+    index = 1
+    while 1:
+        match = BUG_REGEX.search(s,index)
+        if not match:
+            break
+        start = match.start()
+        end = match.end()
+        substring = s[start:end]
+        if substring[0] == '#':  # this of the form "#1234"
+            bugid = substring[1:]
+        else: # this is of the form "bug 1234"
+            bugid = substring[4:]
+        url = '<a href="%s%s">%s</a>' % (BUG_URL,bugid,substring)
+        (s,index) = mstring.replace_sub(s,url,start,end-1)
+    return s
 
 def changelog(filename):
-	try:
-		#print filename
-		fp = open(filename,'r')
-	except IOError:
-		return ""
+    try:
+        #print filename
+        fp = open(filename,'r')
+    except IOError:
+        return ""
 
 
-	s = ""
-	# find first line that isn't blank or a comment
-	while True:
-		line = fp.readline()
-		if not line: break
-		#print line
-		if line[0] not in ['#','','\n']:
-			s = s + line
-			break
-	
-	# append next strings until you reach next "*"
-	while True:
-		line = fp.readline()
-		#print repr(line)
-		if not line or line[0] == '*': break
-		else: s= s + line
-	
-	return s
+    s = ""
+    # find first line that isn't blank or a comment
+    while True:
+        line = fp.readline()
+        if not line: break
+        #print line
+        if line[0] not in ['#','','\n']:
+            s = s + line
+            break
+    
+    # append next strings until you reach next "*"
+    while True:
+        line = fp.readline()
+        #print repr(line)
+        if not line or line[0] == '*': break
+        else: s= s + line
+    
+    return s
diff --git a/src/packages/daily.py b/src/packages/daily.py
index c1746edaae..482dd0bcd3 100644
--- a/src/packages/daily.py
+++ b/src/packages/daily.py
@@ -18,7 +18,7 @@ def error():
 	'<tr><td class="fields">Error in request</td></tr><br>\n'
 	'<tr><td class="item"><img src="%s/?category=generic" align="right" alt="">'
 	'<p>An error was encountered processing your request.  Request a '
-	'different page or check the <a href="%s">fresh ebuilds main page</a>.'
+	'different page or check the <a href="%s">packages.gentoo.org main page</a>.'
 	'</p></td></tr>'
 	'</table>'
 	'</div>') % (config.ICONS,config.FEHOME)
diff --git a/src/packages/ebuilddb.py b/src/packages/ebuilddb.py
index c6d0453de0..3d1040c362 100644
--- a/src/packages/ebuilddb.py
+++ b/src/packages/ebuilddb.py
@@ -8,7 +8,15 @@ import re
 import changelogs
 import MySQLdb
 
-FINDVER = re.compile('-[0-9]')
+# import portage, but temporarily redirect stderr
+if 'portage' not in dir():
+    null = open('/dev/null', 'w')
+    tmp = sys.stderr
+    sys.stderr = null
+    sys.path = ["/usr/lib/portage/pym"]+sys.path
+    import portage
+    sys.stderr = tmp
+    sys.path = sys.path[1:]
 
 def db_connect():
     return MySQLdb.connect(host = config.HOST,
@@ -31,9 +39,14 @@ def parse_ebuild(s):
     s=s.split('/')
     parsed['category'] = s[-3]
     package = s[-1].split('.ebuild')[0]
-    pos=FINDVER.search(package).start()
-    parsed['name'] = package[:pos]
-    parsed['version'] = package[pos+1:]
+    pieces = portage.pkgsplit(package)
+    if not pieces:
+        return None
+    parsed['name'] = pieces[0]
+    if pieces[2] == 'r0':
+        parsed['version'] = pieces[1]
+    else:
+        parsed['version'] = '-'.join(pieces[1:])
 
     return parsed
 
@@ -66,6 +79,10 @@ def create_ebuild_record(db,ebinfo):
 def update_ebuild_record(db,ebinfo):
     c = db.cursor()
     escaped = dict([(x,MySQLdb.escape_string(y)) for (x,y) in ebinfo.items()])
+    query="""REPLACE INTO package VALUES ("%(category)s","%(name)s",\
+        "%(homepage)s","%(description)s","%(license)s");""" % escaped
+    c.execute(query)
+    
     query = ('UPDATE ebuild '
         'SET when_found="%(time)s",'
         'arch="%(archs)s",'
@@ -74,7 +91,13 @@ def update_ebuild_record(db,ebinfo):
         'WHERE category="%(category)s" '
         'AND name="%(name)s" '
         'AND version="%(version)s" ' % escaped)
-    c.execute(query)
+    try:
+        c.execute(query)
+    except MySQLdb.MySQLError, data:
+        print 'error occurred: '
+        print 'query: %s' % query
+        print 'error: %s' % data
+        
 
 def get_extended_info(ebuild):
     filename = os.path.join(config.PORTAGE_DIR,'metadata/cache',
@@ -116,9 +139,8 @@ def main():
     db = db_connect()
     
     for s in ebuilds:
-        try:
-            fields = parse_ebuild(s)
-        except:
+        fields = parse_ebuild(s)
+        if not fields:
             continue
         result = get_ebuild_record(db,fields)
         fields = get_extended_info(fields)
@@ -134,7 +156,6 @@ def main():
             # keywords change, update db
             fields['prevarch'] = result[4]
             update_ebuild_record(db,fields)
-
         
 if __name__ == '__main__':
     main()
diff --git a/src/packages/genrsslist.py b/src/packages/genrsslist.py
index 8bd05978f4..2f7aacc3ea 100644
--- a/src/packages/genrsslist.py
+++ b/src/packages/genrsslist.py
@@ -3,8 +3,8 @@
 import config
 FEHOME = config.FEHOME
 
-print """<div class="centerpage"><H3>Fresh Ebuilds RSS Feeds</H3>
-<p>Fresh Ebuilds provides the following RSS feeds.   Note that some
+print """<div class="centerpage"><H3>packages.gentoo.org RSS Feeds</H3>
+<p>packages.gentoo.org provides the following RSS feeds.   Note that some
 feed readers may have problems with the embedded HTML feeds.</p>
 </div>
 <br>
diff --git a/src/packages/gentoo.py b/src/packages/gentoo.py
index c97fa6a48a..ae8a403e7d 100644
--- a/src/packages/gentoo.py
+++ b/src/packages/gentoo.py
@@ -2,12 +2,13 @@
 """These functions mainly take ebuild info (grabbed from the database and
     convert it to HTML.  See the "main" function at the bottom."""
 
-__revision__ = "$Revision: 1.6 $"
+__revision__ = "$Revision: 1.7 $"
 # $Source: /var/cvsroot/gentoo/src/packages/gentoo.py,v $
 
 import config
 import os
 import time
+import string
 import sys
 import ebuilddb
 import bugs
@@ -96,6 +97,7 @@ def package_to_html(pkginfo, db):
     #bug_string = ('<br><h3>Related bugs:</h3>\n%s' 
     #    % bugs_to_html(pkginfo['name']))
     general = '<tr><td>%s</td></tr>' % general_info_to_html(pkginfo)
+    similar = '<tr><td>%s</td></tr>' % create_similar_pkgs_link(pkginfo)
     table_end = '</table>'
     rows = '\n\t'.join([name, description, releases, general])
     return '\n\t'.join([table_begin, rows, table_end])
@@ -105,10 +107,12 @@ def archs_to_html(ebuilds, heading = None):
     heading = heading or '&nbsp;'
     table_begin = '<table class="releases">'
     header_row = ''.join(['<tr><td><b>%s</b></td>' % heading] +
-        ['<th class="arch">%s</th>' % i for i in config.ARCHLIST] +
+        ['<th class="arch">%s</th>' % i.replace('-',' ') for i in config.ARCHLIST] +
         ['</tr>']
     )
     rows = []
+    ebuilds.sort(cmp_ebuilds)
+    ebuilds.reverse()
     for ebuild in ebuilds:
         masked = is_masked(ebuild)
         archs = ebuild['arch'].split()
@@ -197,7 +201,9 @@ def general_info_to_html(pkg):
     license = ('<td class="license">%s</td>' 
         % license_to_html(pkg['license']))
     changelog = ('<td class="changelog" rowspan="2">'
-        '<a href="%s">ChangeLog</a><td>' % changelogurl)
+        '<a href="%s">ChangeLog</a></td>' % changelogurl)
+    similar = ('<td class="similar" rowspan="2">'
+        '%s</td>' % create_similar_pkgs_link(pkg))
     
     return '\n\t'.join(['<table class="general_info">',
         '<tr>',
@@ -205,6 +211,7 @@ def general_info_to_html(pkg):
         homepage,
         license_header,
         changelog,
+        similar,
         '</tr>',
         '<tr>',
         category,
@@ -212,6 +219,29 @@ def general_info_to_html(pkg):
         '</tr>',
         '</table>'])
         
+def create_similar_pkgs_link(pkg):
+    """Create a link to similar packages"""
+    
+    def strip_chars(mystring):
+        newstring = ''
+        for char in mystring:
+            if char not in string.punctuation:
+                newstring = newstring + char
+            else:
+                newstring = newstring + ' '
+        return newstring
+        
+    description = strip_chars(pkg['description'].lower())
+    
+    words = [word for word in description.split() 
+        if word and len(word)>2 and word not in config.EXCLUDED_FROM_SIMILAR]
+    words = words[:config.SIMILAR_MAX_WORDS] + [pkg['name']]
+    #query = ['[[:<:]]%s[[:>:]].*' % word for word in words]
+    query = ['[[:<:]]%s.*' % word for word in words]
+    query = '(%s){%s,}' % ('|'.join(query), config.SIMILAR_MIN_MATCHES)
+    url = '%ssearch/?sstring=%s' % (config.FEHOME, escape(query))
+    return '<a href="%s">Similar Packages</a>' % url
+    
 def bugs_to_html(package):
     """Given package name (no version #s), return html text of bugs as
     reported by bugzilla"""
@@ -280,13 +310,19 @@ def get_recent_releases(pkg, db, max=config.MAX_RECENT_RELEASES):
     results = c.fetchall()
     #print results
     return [ query_to_dict(i) for i in results ]
-    
+
+def cmp_ebuilds(a, b):
+    """Compare two ebuilds"""
+    fields_a = portage.pkgsplit('%s-%s' % (a['name'], a['version']))
+    fields_b = portage.pkgsplit('%s-%s' % (b['name'], b['version']))
+    return portage.pkgcmp(fields_a, fields_b)
+
 def ebuilds_to_rss(fp, ebuilds, simple=False, subtitle=""):
     """write out ebuild info to RSS file (fp)"""
     fp.write("""<?xml version="1.0" encoding="iso-8859-1"?>
         <rss version="0.92">
         <channel>
-        <title>Fresh ebuilds %s</title>
+        <title>packages.gentoo.org [ %s ]</title>
         <link>%s</link>
         <description>Latest ebuilds from the Gentoo Linux portage tree</description>
         <![CDATA[<link rel="stylesheet" href="%s" type="text/css" title="styled" />]]>
diff --git a/src/packages/query_ebuild.py b/src/packages/query_ebuild.py
index b64681bfbc..e6fb46785d 100644
--- a/src/packages/query_ebuild.py
+++ b/src/packages/query_ebuild.py
@@ -22,9 +22,12 @@ if 0:
 else:
     # let's try the database
     # connect
-    pos=ebuilddb.FINDVER.search(ebuild).start()
-    name = ebuild[:pos]
-    version = ebuild[pos+1:]
+    pieces = gentoo.portage.pkgsplit(ebuild)
+    name = pieces[0]
+    if pieces[2] == 'r0':
+        version = pieces[1]
+    else:
+        version = '-'.join(pieces[1:])
     db = ebuilddb.db_connect()
     # query
     query = ('SELECT ebuild.category,ebuild.name,version,when_found,'
diff --git a/src/packages/query_package.py b/src/packages/query_package.py
index f78af3d940..c30f175cdc 100644
--- a/src/packages/query_package.py
+++ b/src/packages/query_package.py
@@ -11,6 +11,16 @@ from MySQLdb import escape_string
 
 sys.stdout.write('Content-type: text/html; charset=iso-8859-1\n\n')
 
+def cmp_results(a, b):
+    """Compare results a and b"""
+    category_a = a[0]
+    category_b = b[0]
+    
+    order = cmp(category_a, category_b)
+    if order != 0:
+        return order
+        
+    
 def query_to_dict(q):
     pkginfo = {}
     keys = ('category','name','homepage','description','license')
@@ -75,5 +85,5 @@ else:
         'align="right" alt=""> <p>Information on the package you requested'
         ' could not be found.  Be sure to check the'
         ' <a HREF="%s">'
-        'fresh ebuilds main page</a>.</p></td></tr></table>\n'
+        'packages.gentoo.org main page</a>.</p></td></tr></table>\n'
         '</div>' % (config.ICONS,config.FEHOME))
diff --git a/src/packages/search.py b/src/packages/search.py
index 267c56741b..bc118ca12f 100644
--- a/src/packages/search.py
+++ b/src/packages/search.py
@@ -32,7 +32,10 @@ def query_to_dict(q):
 	return pkginfo
 
 def write_to_cache(s):
-	open(cachefile,'w').write(s)
+    try:
+        open(cachefile,'w').write(s)
+    except IOError:
+        pass
 
 def sort_by_weight(a, b):
     """Right now we just sort based on whether the sstring is in the name"""
@@ -59,7 +62,7 @@ if os.path.exists(cachefile):
 escaped = escape_string(sstring)
 query = ('SELECT category,name,homepage,description,license '
 	'FROM package WHERE name REGEXP "%s" or description REGEXP "%s" '
-	'LIMIT %s,%s' 
+	'ORDER BY category LIMIT %s,%s' 
 	% (escaped,escaped,offset,config.MAXPERPAGE))
 
 import ebuilddb
@@ -82,7 +85,7 @@ if not results:
 	'align="right" alt=""> <p>I could not find any ebuild that match'
 	' your query.  Try a different query or check the'
 	' <a HREF="%s">'
-	'fresh ebuilds main page</a>.</p></td></tr></table>\n'
+	'packages.gentoo.org main page</a>.</p></td></tr></table>\n'
 	'</div>\n' % (config.ICONS,config.FEHOME))
 	sys.stdout.write(s)
 	write_to_cache(s)