initial xml2man script

1 files changed, 196 insertions, 0 deletions

diff --git a/docbook2nroff.py b/docbook2nroff.py
new file mode 100755
index 0000000..ae35773
--- /dev/null
+++ b/docbook2nroff.py
@@ -0,0 +1,196 @@
+#!/usr/bin/env python 
+import sys
+from xml import sax
+
+class Variable(object):
+    def __init__(self):
+        self.term = None
+        self.listitem = None
+
+    def __str__(self):
+        return "%s %s" % (self.term, self.listitem)
+
+class TextBlock(object):
+    def __init__(self, text = "", order = 0):
+        self.text = text
+        self.order = order
+        
+    def __str__(self):
+        return self.text
+        
+class Para(TextBlock):
+    def __str__(self):
+        return "\n.PP\n" + self.text.lstrip()
+
+class Code(TextBlock):
+    def __str__(self):
+        return "\n.nf\n%s\n.fi\n" % self.text.lstrip()
+
+class DocBookToNroffHandler(sax.ContentHandler):
+    def __init__(self, out=sys.stdout):
+        ###########################################################################################
+        self.out = out
+        self.stack = []
+        ###########################################################################################
+        
+        self.refentryinfo = {}
+        self.refmeta = {}
+        self.refnamediv = {}
+        self.synopsis = {}
+        self.refsection = {}
+
+        self.mypara = None
+        self.myvar = None
+        self.myprogramlisting = None
+        self.elementcount = 0
+        self.formatting = []
+        self.justformatted = False
+
+    def is_formatting(self, what):
+        return what in [ "command", "option", "parameter", "replaceable", "filename", "emphasis" ]
+
+    def translate_formatting(self, what):
+        if what in ["command", "option", "parameter", "filename"]:
+            return "B"
+        elif what in ["emphasis", "replaceable"]:
+            return "I"
+        return ""
+        
+    def startElement(self, name, attrs):
+        ###########################################################################################
+        #print >> self.out, "+" + name
+        name = name.lower()    
+        self.stack.insert(0, name)
+        self.elementcount += 1
+        #print >> self.out, self.stack
+        ###########################################################################################
+#        if name == "arg":
+#            attrs.get("choice")
+
+        if "refsection" == name:
+            self.refsection["id"] = attrs["id"]
+
+        if "para" == name:
+            self.mypara = Para(order=self.elementcount)
+        
+        if "varlistentry" == name:
+            self.myvar = Variable()
+
+        if "programlisting" == name:
+            self.myprogramlisting = Code(order=self.elementcount)
+
+        if self.is_formatting(name):
+            self.formatting.append(self.translate_formatting(name))
+
+    def characters(self, content):    
+        ###########################################################################################
+        #print >> self.out, content
+        ###########################################################################################
+        
+        if "refentryinfo" in self.stack:
+            if "refentrytitle" in self.stack:
+                self.refentryinfo["refentrytitle"] = content
+            if "manvolnum" in self.stack:
+                self.refentryinfo["manvolnum"] = content
+            if "productname" in self.stack:
+                self.refentryinfo["productname"] = content
+            if "date" in self.stack:
+                self.refentryinfo["date"] = content
+        
+        elif "refmeta" in self.stack:
+            if "refmeta" in self.stack:
+                self.refmeta["manvolnum"] = content
+        
+        elif "refnamediv" in self.stack:
+            if "refname" in self.stack:
+                self.refentryinfo["refname"] = content
+            if "refpurpose" in self.stack:
+                self.refentryinfo["refpurpose"] = content
+
+        elif "cmdsynopsis" in self.stack:
+            if "command" in self.stack:
+                self.synopsis["command"] = content
+            elif "arg" in self.stack:
+                if not self.synopsis.has_key("args"):
+                    self.synopsis["args"] = []
+                self.synopsis["args"].append(content) 
+
+        elif "refsection" in self.stack:
+            if "title" in self.stack:
+                self.refsection["title"] = content
+            elif "varlistentry" in self.stack:
+                if "term" in self.stack:
+                    self.myvar.term = content
+                elif "listitem" in self.stack:
+                    self.myvar.listitem = content
+            elif "programlisting" in self.stack:
+                self.myprogramlisting.text += content
+            elif "para" in self.stack:
+                if self.formatting:
+                    self.mypara.text += "\n.%s %s\n" % ("".join(self.formatting), content)
+                    self.justformatted = True
+                else:
+                    if self.justformatted:
+                        self.mypara.text += content.lstrip()
+                    else:
+                        self.mypara.text += content
+                    self.justformatted = False
+    
+    
+    def endElement(self, name):
+        ###########################################################################################
+        #print >> self.out, "-" + name
+        name = name.lower()
+        del self.stack[0]
+        #print >> self.out, self.stack
+        ###########################################################################################
+        
+        if "refentryinfo" == name:
+            print >> self.out, ".TH %(refentrytitle)s %(manvolnum)s \"%(date)s\" \"USER COMMANDS\"" % self.refentryinfo
+
+        if "refnamediv" == name:
+            print >> self.out, ".SH NAME \n %(refname)s \- %(refpurpose)s" % self.refentryinfo
+
+        if "cmdsynopsis" == name:
+            self.synopsis["argcombo"] = " ".join(self.synopsis["args"])
+            print >> self.out, ".SH Synopsis \n.B %(command)s \n%(argcombo)s" % self.synopsis
+
+        if "refsection" == name:
+            print >> self.out, ".SH %(title)s \n" % self.refsection
+            if self.refsection.has_key("varlist"):
+                for v in self.refsection["varlist"]:
+                    print >> self.out, ".TP \n.B %s\n%s" % (v.term,v.listitem)
+
+            if self.refsection.has_key("textblock"):
+                self.refsection["textblock"].sort(cmp = lambda a,b: a.order - b.order)
+                for p in self.refsection["textblock"]:
+                    print >> self.out, str(p)
+
+            self.refsection = {}
+
+        if "varlistentry" == name:
+            if not self.refsection.has_key("varlist"):
+                self.refsection["varlist"] = []
+            self.refsection["varlist"].append(self.myvar)
+            self.myvar = None
+
+        if "programlisting" == name:
+            if not self.refsection.has_key("textblock"):
+                self.refsection["textblock"] = []
+            self.refsection["textblock"].append(self.myprogramlisting)
+            self.myprogramlisting = None
+
+        if "para" == name:
+            if not self.refsection.has_key("textblock"):
+                self.refsection["textblock"] = []
+            self.refsection["textblock"].append(self.mypara)
+            self.mypara = None
+
+        if self.is_formatting(name):
+            self.formatting.pop()
+
+
+
+parser = sax.make_parser()
+parser.setContentHandler(DocBookToNroffHandler())
+parser.parse(open(sys.argv[1], "r"))