CSV: Added optional Regexp check for fieldnames.

lib/modules/Parser.pmod/CSV.pike

@@ -13,23 +13,32 @@
 inherit Parser.Tabular;
 
 //! This function consumes the header-line preceding a typical comma,
-//! semicolon or tab separated value list.
+//! semicolon or tab separated value list and autocompiles a format
+//! description from that.  After this function has
+//! successfully parsed a header-line, you can proceed with
+//! either @[fetchrecord()] or @[fetch()] to get the remaining records.
 //!
 //! @param delimiters
 //! Explicitly specify a string containing all the characters that should
-//! be considered field delimiters.  If not specified, the function will
-//! try to autodetect the single delimiter in use.
+//! be considered field delimiters.  If not specified or empty, the function
+//! will try to autodetect the single delimiter in use.
+//!
+//! @param matchfieldname
+//! A string containing a regular expression, using @[Regexp.SimpleRegexp]
+//! syntax, or an object providing a @[Regexp.SimpleRegexp.match()]
+//! single string argument compatible method, that must match all the
+//! individual fieldnames before the header will be considered valid.
 //!
 //! @returns
 //! It returns true if a CSV head has successfully been parsed.
 //!
 //! @seealso
-//!  @[fetchrecord()], @[compile()]
-int parsehead(void|string delimiters)
+//!  @[fetchrecord()], @[fetch()], @[compile()]
+int parsehead(void|string delimiters,void|string|object matchfieldname)
 { if(skipemptylines())
     return 0;
-  string line=_in->gets();
-  if(!delimiters)
+  { string line=_in->gets();
+    if(!delimiters||!sizeof(delimiters))
     { int countcomma,countsemicolon,counttab;
       countcomma=countsemicolon=counttab=0;
       foreach(line;;int c)
@@ -45,14 +54,36 @@ int parsehead(void|string delimiters)
        countsemicolon>counttab?";":"\t";
     }
     _in->unread(line+"\n");
+  }
+
   multiset delim=(<>);
   foreach(delimiters;;int c)
     delim+=(<c>);
+
   array res=({ (["single":1]),0 });
   mapping m=(["delim":delim]);
+
+  if(!objectp(matchfieldname))
+    matchfieldname=Regexp(matchfieldname||"");
   _eol=0;
-  do res+=({m+(["name":_getdelimword(m)])});
+  if(mixed err = catch
+    { _checkpoint checkp=_checkpoint();
+      do
+      { string field=_getdelimword(m);
+        res+=({ m+(["name":field]) });
+	if(String.width(field)>8)
+	  field=string_to_utf8(field);	  // FIXME dumbing it down for Regexp()
+        if(!matchfieldname->match(field))
+	  throw(1);
+      }
       while(!_eol);
+    })
+    switch(err)
+    { default:
+	throw(err);
+      case 1:
+	return 0;
+    }
   setformat( ({res}) );
   return 1;
 }

lib/modules/Parser.pmod/Tabular.pike

@@ -112,7 +112,7 @@ private string gets(int n)
   return s;
 }
 
-private class checkpoint
+class _checkpoint
 { private string oldalread;
 
   void create()
@@ -428,12 +428,12 @@ mapping fetch(void|array|mapping format)
 ret:
   { if(arrayp(format))
     { mixed err=catch
-      { checkpoint checkp=checkpoint();
+      { _checkpoint checkp=_checkpoint();
         foreach(format;;array|mapping fmt)
           if(arrayp(fmt))
             for(int found=0;;found=1)
             { mixed err=catch
-              { checkpoint checkp=checkpoint();
+              { _checkpoint checkp=_checkpoint();
 		mapping rec=getrecord(fmt,found);
 	        foreach(rec;string name;mixed value)
                   add2map(ret,name,value);
@@ -468,6 +468,8 @@ ret:
     { int found;
       do
       { found=0;
+	if(!mappingp(format))
+	  error("Empty format definition\n");
         foreach(format;string name;array|mapping subfmt)
           for(;;)
 	  { if(verb<0)