Add translation to support lg2mml for infty

bin/lg2mml

@@ -24,4 +24,4 @@ fi
 
 BNAME=`basename $1 .lg`
 
-python $LgEvalDir/src/lg2txt.py $1 $LgEvalDir/translate/mathMLMap.csv > $BNAME.mml
+python $LgEvalDir/src/lg2txt.py $1 $LgEvalDir/translate/mathMLMap.csv $LgEvalDir/translate/infty_to_crohme.csv > $BNAME.mml

src/lg2txt.py

@@ -11,11 +11,41 @@
 # Copyright (c) 2012-2014 Richard Zanibbi and Harold Mouchere
 ################################################################
 
+import os
+import re
 import sys
 import csv
 from bs4 import BeautifulSoup
 from lg import *
 
+def readtranslateFile(mapFile):
+    """Read in symbol and structure mappings from a file."""
+    try:
+        fileReader = csv.reader(open(mapFile))
+    except:
+        sys.stderr.write('  !! IO Error (cannot open): ' + mapFile + '\n')
+        return
+
+    symbolsMap = {}
+    relationsMap = {}
+    readingSymbols = True
+    for row in fileReader:
+        # Skip blank lines and comments.
+        if len(row) == 0:
+            continue
+        elif row[0].strip()[0] == '#':
+            continue
+        elif row[0].strip() == 'SYMBOLS':
+            readingSymbols = True
+        elif row[0].strip() == 'RELATIONSHIPS':
+            readingSymbols = False
+        else:
+            if readingSymbols:
+                symbolsMap[row[0]] = row[1]
+            else:
+                relationsMap[row[0]] = row[1]
+    return symbolsMap, relationsMap
+
 def readMapFile(fileName):
     """Read in symbol and structure mappings from a file."""
     try:
@@ -286,33 +316,7 @@ def cleanRows( mmlFile ):
 
     return rowSoup.prettify()
 
-
-# def postprocess(mml_out):
-#     # Add linebreaks for tags expect for mi, mo, mn
-#     mml_out = mml_out.replace("><", ">\n<")
-#     mml_out = mml_out.replace("</", "\n</")
-#     mml_out = mml_out.replace("\n</mi", "</mi")
-#     mml_out = mml_out.replace("\n</mo", "</mo")
-#     mml_out = mml_out.replace("\n</mn", "</mn")
-#     # tags with no indentation
-#     constant_tags = ["mi", "mo", "mn"]
-#     # Counts number of tab sapces
-#     tab = 0
-#     out = []
-#     for line in mml_out.split("\n"):
-#         if not line:
-#             continue
-# #     if re.match(r"<\w", line):
-#         if line.startswith("<m"):
-#             out.append(tab * "\t" + line)
-#             if not line[1:3] in constant_tags:
-#                 tab += 1
-#         elif line.startswith("</"):
-#             tab -= 1
-#             out.append(tab * "\t" + line)
-#     return "\n".join(out)
-
-def main(lg_file, mapFile):
+def lg2mml(lg_file, mapFile):
 
     lg = Lg(lg_file)
 
@@ -360,18 +364,52 @@ def main(lg_file, mapFile):
     mml_out = cleanRows("\n".join(mml_out_raw)) 
     return mml_out
 
+def preprocess(filename, translateFile, translate=True):
+    lg = Lg(filename)
+    print(lg)
+    lg_NE_string = lg.csv()
+    if translate:
+        symbolsMap, relationsMap = readtranslateFile(translateFile)
+        edge_pattern = re.compile(r'^E,')
+        node_pattern = re.compile(r'(^N,)|#')
+        symbol_temp = relabel(lg_NE_string, symbolsMap, node_pattern)
+        edge_temp = relabel(lg_NE_string, dict(**symbolsMap, **relationsMap), 
+                            edge_pattern)
+        lg_NE_string = symbol_temp + "\n" + edge_temp
+    temp_file = "temp.lg"
+    with open(temp_file, 'w') as f:
+        f.writelines(lg_NE_string)
+    return temp_file
+        
+def relabel(lg_NE_string, Map, pattern):
+    temp = '\n'.join([line for line in lg_NE_string.split('\n') 
+                       if re.match(pattern, line)])
+    for source_label, mapped_label in Map.items():
+        temp = temp.replace("," + source_label + ",", "," + mapped_label + ",")
+    return temp
+
+def main(lg_file, mapFile, translateFile):
+    temp_lg_file = preprocess(lg_file, translateFile)
+    mml_out = lg2mml(temp_lg_file, mapFile)
+    os.remove(temp_lg_file)
+    return mml_out
+
 if __name__ == '__main__':
     if len(sys.argv) < 2:
-        print("Usage: [[python]] lg2txt.py <infile.lg> [mapfile.csv]")
+        print("Usage: [[python]] lg2txt.py <infile.lg> [mapfile.csv] [translatefile.csv]")
         print("")
         print("   Produces a text file for label graph file")
         print("   <infile.lg>. A symbol and structure map file (mapfile.csv)")
         print("   may be provided to override default (latex) mappings.")
         sys.exit()
     lg_file = sys.argv[1]
-    if len(sys.argv) > 2:
+    if len(sys.argv) > 3:
+        mapFile = sys.argv[2]
+        translateFile = sys.argv[3]
+    elif len(sys.argv) > 2:
         mapFile = sys.argv[2]
+        translateFile = None
     else:
         mapFile = None
-    mml_out = main(lg_file, mapFile)
+    mml_out = main(lg_file, mapFile, translateFile)
     print(mml_out)

translate/infty_to_crohme.csv

@@ -177,4 +177,4 @@ UNDER,Below
 LSUB,Sub
 UPPER,Above
 LSUP,Sup
-PUNC,Right
+PUNC,Sub