From da7359cd452f2ded9e05e753fb125508343b8587 Mon Sep 17 00:00:00 2001
From: Thibaut Horel <thibaut.horel@gmail.com>
Date: Thu, 4 Feb 2016 20:00:45 -0500
Subject: Split sort utilities

---
 sortutils.py | 138 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 138 insertions(+)
 create mode 100644 sortutils.py

(limited to 'sortutils.py')

diff --git a/sortutils.py b/sortutils.py
new file mode 100644
index 0000000..419fe03
--- /dev/null
+++ b/sortutils.py
@@ -0,0 +1,138 @@
+import config
+import copy
+from utils import txtize
+from entry import buildAuthorTable
+import re
+
+# List: must map from month number to month name.
+MONTHS = [None, "January", "February", "March", "April", "May", "June",
+          "July", "August", "September", "October", "November", "December"]
+
+
+def splitEntriesBy(entries, field):
+    """Take a list of BibTeX entries and the name of a bibtex field; return
+       a map from vield value to list of entry."""
+    result = {}
+    for ent in entries:
+        key = ent.get(field)
+        if field in config.MULTI_VAL_FIELDS:
+            key = [k.strip() for k in key.split(',')]
+        else:
+            key = [key]
+        for k in key:
+            try:
+                result[k].append(ent)
+            except:
+                result[k] = [ent]
+    return result
+
+
+def splitSortedEntriesBy(entries, field):
+    """Take inputs as in splitEntriesBy, where 'entries' is sorted by 'field'.
+       Return a list of (field-value, entry-list) tuples, in the order
+       given in 'entries'."""
+    result = []
+    curVal = "alskjdsakldj"
+    curList = []
+    for ent in entries:
+        key = ent.get(field)
+        if key == curVal:
+            curList.append(ent)
+        else:
+            curVal = key
+            curList = [ent]
+            result.append((curVal, curList))
+    return result
+
+
+def sortEntriesBy(entries, field, default):
+    """Take inputs as in splitEntriesBy, and return a list of entries sorted
+       by the value of 'field'. Entries without 'field' are sorted as if their
+       value were 'default'.
+       """
+    tmp = []
+    i = 0
+    for ent in entries:
+        i += 1
+        v = ent.get(field, default)
+        if v.startswith("<span class='bad'>"):
+            v = default
+        if field in config.MULTI_VAL_FIELDS:
+            for v_j in v.split(','):
+                ent_j = copy.deepcopy(ent)
+                ent_j.__setitem__(field, v_j.strip())
+                tmp.append((txtize(v_j.strip()), i, ent_j))
+        else: tmp.append((txtize(v), i, ent))
+    tmp.sort()
+    return [t[2] for t in tmp]
+
+
+def splitEntriesByAuthor(entries):
+    """Take a list of entries, sort them by author names, and return:
+         a sorted list of (authorname-in-html, bibtex-entry-list) tuples,
+         a map from authorname-in-html to name-for-url.
+       Entries with multiple authors appear once per author.
+    """
+    collapsedAuthors = buildAuthorTable(entries)
+    entries = sortEntriesByDate(entries)
+    result = {} # Name in sorting order -> entries
+    htmlResult = {} # name in sorting order -> Full name
+    url_map = {} # Full name -> Url
+    for ent in entries:
+        for a in ent.parsedAuthor:
+            canonical = collapsedAuthors[a]
+            url = canonical.getHomepage()
+            sortkey = canonical.getSortingName()
+            secname = canonical.getSectionName()
+            if url:
+                url_map[secname] = url
+
+            htmlResult[sortkey] = secname
+            result.setdefault(sortkey, []).append(ent)
+    sortnames = result.keys()
+    sortnames.sort()
+    sections = [ (htmlResult[n], result[n]) for n in sortnames ]
+    return sections, url_map
+
+## def sortEntriesByAuthor(entries):
+##     tmp = []
+##     i = 0
+##     for ent in entries:
+##         i += 1
+##         authors = [ txtize(" ".join(a.von+a.last+a.first+a.jr))
+##                     for a in ent.parsedAuthor ]
+##         tmp.append((tuple(authors), i, ent))
+##     tmp.sort()
+##     return [ t[2] for t in tmp ]
+
+def sortEntriesByDate(entries):
+    """Sort a list of entries by their publication date."""
+    tmp = []
+    i = 0
+    for ent in entries:
+        i += 1
+        if (ent.get('month') == "forthcoming" or
+            ent.get('year') == "forthcoming"):
+            tmp.append((20000*13, i, ent))
+            continue
+        try:
+            monthname = ent.get("month")
+            if monthname is not None:
+                match = re.match(r"(\w+)--\w+", monthname)
+                if match:
+                    monthname = match.group(1)
+            mon = MONTHS.index(monthname)
+        except ValueError:
+            print "Unknown month %r in %s"%(ent.get("month"), ent.key)
+            mon = 0
+
+        try:
+            date = int(ent['year'])*13 + mon
+        except KeyError:
+            print "ERROR: No year field in %s"%ent.key
+            date = 10000*13
+        except ValueError:
+            date = 10000*13
+        tmp.append((date, i, ent))
+    tmp.sort()
+    return [ t[2] for t in tmp ]
-- 
cgit v1.2.3-70-g09d2