Wicri:Dilib source, module Nlm, commande NlmPmc2Wicri

Code source

/*   -*- coding: utf-8 -*-  */

/*
           NlmPmc2Wicri.c

 */

#include <stdio.h>     /* for printf */
#include <stdlib.h>    /* for exit */
#include <getopt.h>
#include "SxmlNode.h"

void proceedText(SxmlNode *textNode)
{
  SxmlNode *elemText;
  SxmlReset (textNode);
  while ((elemText=SxmlNextNode(textNode)))
    {
      if (SxmlIsText(elemText))
	{
	  SxmlPrint(elemText);
	  continue;
	}
      if(SxmlNodeHasName(elemText,"sup")) 
	 {
	   printf ("<sup>");
	   proceedText(elemText);
	   printf ("</sup>");
	   continue;
	 }
     if(SxmlNodeHasName(elemText,"xref")) 
	 {
	   char *rid;
	   rid=SxmlGetAttribute(elemText,"rid");
	   printf ("<span id=\"pmc xref %s\">[[#pmc %s|", rid, rid);
	   proceedText(elemText);
	   printf ("]]</span>");
	   continue;
	 }
    }
}

void proceedTitle(SxmlNode *titleNode, int level)
{
  switch(level)
    {
    case 1: printf("=="); break;
    case 2: printf("==="); break;
    case 3: printf("===="); break;
    }
  proceedText(titleNode);
  switch(level)
    {
    case 1: printf("==\n"); break;
    case 2: printf("===\n"); break;
    case 3: printf("====\n"); break;
    }
}

void proceedP(SxmlNode *parNode)
{
  proceedText(parNode);
  printf("\n\n");
}

void proceedSec(SxmlNode *secNode, int level)
{
  SxmlNode *titleNode;
  SxmlNode *partSec;
  int tagSec1;
  
  tagSec1=0;
  SxmlReset (secNode);
  
  while ((partSec=SxmlNextNode(secNode)))
    {
       if(SxmlNodeHasName(partSec,"title")) 
	 {
	   proceedTitle(partSec, level);
	   continue;
	 }
       if (level==1) 
	 {
	   if (tagSec1==0)
	     {
	       printf ("{{Begin article body}}\n");
	       tagSec1=1;
	     }
	 }
       if(SxmlNodeHasName(partSec,"sec")) 
	 {
	   proceedSec(partSec, level+1);
	   continue;
	 }    
       if(SxmlNodeHasName(partSec,"p")) 
	 {
	   proceedP(partSec);
	   continue;
	 }    
    }
  if (tagSec1==1) printf ("{{End article body}}\n");
}


void proceedBody(SxmlNode *bodyNode)
{
  SxmlNode *bodyNodeItem;
  SxmlReset (bodyNode);
  while ((bodyNodeItem=SxmlNextNode(bodyNode)))
    {
      if(SxmlNodeHasName(bodyNodeItem,"sec")) proceedSec(bodyNodeItem, 1);
    }
}

void proceedRef(SxmlNode *refItem)
{
  char *id;
  SxmlNode *mixedCit;
  SxmlNode *refItemItem;
  id=SxmlGetAttribute(refItem, "id");
  if (!id)return;
  mixedCit=SxmlGetFirstChildByTagName(refItem, "mixed-citation");
  if (!mixedCit)return;
  
  printf("#");
  printf("<span id=\"pmc %s\">[[#pmc xref %s|↑]]</span>", id, id);
  SxmlReset(mixedCit);
  while ((refItemItem=SxmlNextNode(mixedCit)))
    {
      if (SxmlNodeHasName(refItemItem, "name"))
	{
	  SxmlNode *surnameNode;
	  SxmlNode *givenNameNode;
	  char *surname;
	  char *givenName;
	  surnameNode=SxmlGetFirstChildByTagName(refItemItem,"surname");
	  if (!surnameNode)continue;
	  surname =SxmlLeafText(surnameNode);
	  givenNameNode=SxmlGetFirstChildByTagName(refItemItem,"given-names");
	  if (!givenNameNode)continue;
	  givenName=SxmlLeafText(givenNameNode);
	  printf ("[[Has cited author::%s %s|%s %s]]", givenName, surname, surname, givenName);
	  if (SxmlGetNextSiblingByTagName(refItemItem, "name")) printf(", "); 
	  else printf(". ");
	  continue;
	}
     if (SxmlNodeHasName(refItemItem, "article-title"))
       {
	 printf("%s. ", SxmlLeafText(refItemItem)); 
	  continue;
       }
     if (SxmlNodeHasName(refItemItem, "source"))
       {
	 printf("''%s'', ", SxmlLeafText(refItemItem)); 
	  continue;
       }
    }
  printf("\n");
}

void proceedRefList(SxmlNode *refList)
{
  SxmlNode *refItem;
  SxmlReset(refList);
  while((refItem=SxmlNextNode(refList)))
    {
      if (SxmlNodeHasName(refItem, "ref"))
	proceedRef(refItem);
    }
}

void proceedBack(SxmlNode *backNode)
{
  SxmlNode *backNodeItem;
  SxmlReset (backNode);
  while ((backNodeItem=SxmlNextNode(backNode)))
    {
      if(SxmlNodeHasName(backNodeItem,"sec"))
	{
	  proceedSec(backNodeItem, 1);
	  continue;
	}
      if(SxmlNodeHasName(backNodeItem,"ref-list"))
	{
	  printf("==References==\n");
	  printf ("{{Begin article body}}\n");
	  proceedRefList(backNodeItem);
	  printf ("{{End article body}}\n");
	}
    }
}

void proceedArticle(SxmlNode *articleNode)
{
  SxmlNode *bodyNode;
  SxmlNode *backNode;
  bodyNode=SxmlGetFirstChildByTagName(articleNode, "body");
  if (bodyNode)  proceedBody(bodyNode);
  backNode=SxmlGetFirstChildByTagName(articleNode, "back");
  if (backNode)  proceedBack(backNode);
}

int main(int argc, char **argv)
{
  SxmlNode *docInput;
  while ((docInput=SxmlInputNextDocumentElement()))
    {
      if(SxmlNodeHasName(docInput,"article"))
	{
	  proceedArticle(docInput);
	}
    }
}

Makefile

A insérer dans $DILIB_IMPORT/Nlm/Nlm.mk

                             $(DILIB)/bin/NlmPmc2Wicri \

$(DILIB)/bin/NlmPmc2Wicri: $(DILIB_IMPORT)/Nlm/NlmPmc2Wicri.c \
                                  $(DILIB_TARGET)/Date/SgmlPath.lib
:$(CC)  $(DILIB_IMPORT)/Nlm/NlmPmc2Wicri.c -Wall -I$(DILIB)/include \
	                         $(DILIB)/lib/libDilib.a -o $@

Mise en œuvre

La suite de commande ci-dessous range dans un répertoire importNlm le fichier NlmPmc2Wicri.c et les mises à jour de makefile dans importNlm/NlmUpdate.mk.

WicriGetPage -l wicri-outils.fr -p "Wicri:Dilib source, module Nlm, commande NlmPmc2Wicri"   \
      | MediaWikiExtractSources -w | HfdStoreFile

Simple test

Pour tester la commande en local, faire :

gcc importNlm/NlmPmc2Wicri.c $DILIB_CC -o importNlm/NlmPmc2Wicri
cat  myPmc_resultFile.xml \
 |NlmPmc2Sxml              \
 | Utf8FromHexEntity       \
 | SxmlUnIndent            \
 | importNlm/NlmPmc2Wicri

Voir aussi

• Dilib, module Nlm, commande NlmPmc2Wicri