From 362dd6ee09d7901b748d23e793cbd63a24877e57 Mon Sep 17 00:00:00 2001
From: syntaxaire <syntaxaire@gmail.com>
Date: Wed, 10 Jul 2019 22:02:57 -0400
Subject: [PATCH] Repair of more invalid XML to allow for newlines in
 description output Minor config formatting

---
 config.yml         | 25 +++++++++++++------------
 qud_object_tree.py | 27 +++++++++++++++++++++------
 2 files changed, 34 insertions(+), 18 deletions(-)

diff --git a/config.yml b/config.yml
index ac157d5..d279a70 100644
--- a/config.yml
+++ b/config.yml
@@ -4,21 +4,19 @@ Interface:
   # Interface options
   Initial expansion targets:
     # These object category IDs will be expanded when the tree is initialized:
-    [Food, MeleeWeapon, MissileWeapon, Armor, Shield, Token, LightSource, Tool, Tonic, Trinket, Energy Cell, Security Card, Barathrumite, SapientMutatedFlower]
+    [Food, MeleeWeapon, MissileWeapon, Armor, Shield, Token, LightSource, Tool, Tonic, Trinket,
+     Energy Cell, Security Card, Barathrumite, SapientMutatedFlower]
 
 Templates:
   # Options controlling the generation of wiki infobox templates
   Fields:
     # The fields that (if applicable) will be generated for each item, in this order:
     # (title comes first, but is not listed here)
-    [image, lv, pv, maxpv, vibro, pvpowered, hp, av, dv,
-    ma, tohit, ammo, accuracy, shots, maxammo, maxvol,
-    liquidgen, liquidtype, maxcharge, charge, weight, commerce,
-    complexity, tier, bits, canbuild, skill, colorstr, renderstr, id,
-    bookid, lightradius, hunger, thirst, twohanded, metal,
-    lightprojectile, extra, strength, agility, toughness,
-    intelligence, willpower, ego, acid, electric, cold, heat,
-    desc]
+    [image, lv, pv, maxpv, vibro, pvpowered, hp, av, dv, ma, tohit, ammo, accuracy, shots, maxammo,
+     maxvol, liquidgen, liquidtype, maxcharge, charge, weight, commerce, complexity, tier, bits,
+     canbuild, skill, colorstr, renderstr, id, bookid, lightradius, hunger, thirst, twohanded,
+     metal, lightprojectile, extra, strength, agility, toughness, intelligence, willpower, ego,
+     acid, electric, cold, heat, desc]
   Image overrides:
     # If the image for an object was uploaded under a different name than in the
     # game data, override it here:
@@ -26,6 +24,8 @@ Templates:
 
 Wiki:
   Categories:
+    # Each wiki category is followed by a list of nodes in the inheritance tree whose child objects
+    # should be in that category. Objects may also be explicitly specified.
     Items: [Item]
     Weapons: [MeleeWeapon, MissileWeapon]
     Equipment: [Armor, Shield]
@@ -33,6 +33,7 @@ Wiki:
     Thrown weapons: [BaseThrownWeapon]
     Plants: [Plant]
     Creatures: [Creature]
-    Characters: [Phinae Hoshaiah, Barathrumite, Argyve, ElderBob, Mehmet, Warden Ualraig, Warden Esthers,
-      Eskhind, Meyehind, Liihart, Isahind, Neelahind, Keh, Kesehind, Angohind, Lulihart, AgateSeveranceStar,
-      Mayor Nuntu, Oboroqoru, Asphodel, Yurl, Euclid]
\ No newline at end of file
+    Characters: [Phinae Hoshaiah, Barathrumite, Argyve, ElderBob, Mehmet, Warden Ualraig,
+                 Warden Esthers, Eskhind, Meyehind, Liihart, Isahind, Neelahind, Keh, Kesehind,
+                 Angohind, Lulihart, AgateSeveranceStar, Mayor Nuntu, Oboroqoru, Asphodel, Yurl,
+                 Euclid]
\ No newline at end of file
diff --git a/qud_object_tree.py b/qud_object_tree.py
index 72946b2..d7be207 100644
--- a/qud_object_tree.py
+++ b/qud_object_tree.py
@@ -11,13 +11,28 @@
 
 def load(path):
     """Load ObjectBlueprints.xml from the specified filepath and return a reference to the root."""
-    # Do some repair of invalid XML
-    pattern = re.compile("(&#15;)|(&#11;)")
-    repaired = []
+    # Do some repair of invalid XML:
+    # First, delete some invalid characters
+    pat_invalid = re.compile("(&#15;)|(&#11;)")
     with open(path, 'r', encoding='utf-8') as f:
-        for line in f:
-            repaired.append(pattern.sub('', line))
-    raw = et.fromstringlist(repaired)
+        contents = f.read()
+    contents = re.sub(pat_invalid, '', contents)
+
+    # Second, replace line breaks inside attributes with proper XML line breaks
+    # ^\s*<[^!][^>]*\n[^>]*>
+    pat_linebreaks = r"^\s*<[^!][^>]*\n.*?>"
+    match = re.search(pat_linebreaks, contents, re.MULTILINE)
+    while match:
+        before = match.string[:match.start()]
+        fixed = match.string[match.start():match.end()].replace('\n', '&#10;')
+        after = match.string[match.end():]
+        contents = before + fixed + after
+        match = re.search(pat_linebreaks, contents, re.MULTILINE)
+    # Uncomment to have a diff-able file to double check XML repairs.
+    # with open('test_output.xml', 'w', encoding='utf-8') as f:
+    #     f.write(contents)
+
+    raw = et.fromstring(contents)
 
     # Build the Qud object hierarchy from the XML data
     for element in raw: