parser: Implement xmlCtxtGetInputWindow

See #762.

include/libxml/parser.h

@@ -111,7 +111,11 @@ struct _xmlParserInput {
     const char *directory XML_DEPRECATED_MEMBER;
     /* Base of the array to parse */
     const xmlChar *base;
-    /* Current char being parsed */
+    /**
+     * @deprecated Use #xmlCtxtGetInputWindow
+     *
+     * Current char being parsed
+     */
     const xmlChar *cur;
     /* end of the array to parse */
     const xmlChar *end;
@@ -1949,6 +1953,12 @@ XMLPUBFUN int
 					 int *line,
 					 int *col,
 					 unsigned long *bytePos);
+XMLPUBFUN int
+		xmlCtxtGetInputWindow	(xmlParserCtxt *ctxt,
+					 int inputIndex,
+					 const xmlChar **startOut,
+					 int *sizeInOut,
+					 int *offsetOut);
 XMLPUBFUN void
 		xmlCtxtSetErrorHandler	(xmlParserCtxt *ctxt,
 					 xmlStructuredErrorFunc handler,

parserInternals.c

@@ -3436,13 +3436,60 @@ xmlCtxtGetInputPosition(xmlParserCtxt *ctxt, int inputIndex,
         unsigned long consumed;
 
         consumed = input->consumed;
-        xmlSaturatedAddSizeT(&consumed, input->end - input->base);
+        xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
         *utf8BytePos = consumed;
     }
 
     return 0;
 }
 
+/**
+ * Return window into input data.
+ *
+ * Should only be used by error handlers or SAX callbacks.
+ * The returned pointer is only valid until the callback returns.
+ *
+ * Because of entities, there can be multiple inputs. Non-negative
+ * values of `inputIndex` (0, 1, 2, ...)  select inputs starting
+ * from the outermost input. Negative values (-1, -2, ...) select
+ * inputs starting from the innermost input.
+ *
+ * @since 2.15.0
+ *
+ * @param ctxt  parser context
+ * @param inputIndex  input index
+ * @param startOut  start of window (output)
+ * @param sizeInOut  maximum size of window (in)
+ *                   actual size of window (out)
+ * @param offsetOut  offset of current position inside
+ *                   window (out)
+ * @returns 0 on success, -1 if arguments are invalid
+ */
+int
+xmlCtxtGetInputWindow(xmlParserCtxt *ctxt, int inputIndex,
+                      const xmlChar **startOut,
+                      int *sizeInOut, int *offsetOut) {
+    xmlParserInput *input;
+
+    if (ctxt == NULL || startOut == NULL || sizeInOut == NULL ||
+        offsetOut == NULL)
+        return -1;
+
+    if (inputIndex < 0) {
+        inputIndex += ctxt->inputNr;
+        if (inputIndex < 0)
+            return -1;
+    }
+    if (inputIndex >= ctxt->inputNr)
+        return -1;
+
+    input = ctxt->inputTab[inputIndex];
+
+    xmlParserInputGetWindow(input, startOut, sizeInOut, offsetOut);
+
+    return 0;
+}
+
 /************************************************************************
  *									*
  *		Handling of node information				*

testparser.c

@@ -201,6 +201,75 @@ testInvalidCharRecovery(void) {
     return err;
 }
 
+static void
+testCtxtInputGetterError(void *errCtxt, const xmlError *error) {
+    int *err = errCtxt;
+    xmlParserCtxt *ctxt = error->ctxt;
+    const char *filename;
+    int line, col;
+    unsigned long bytePos;
+    const xmlChar *start;
+    int size, offset;
+
+    xmlCtxtGetInputPosition(ctxt, 0, &filename, &line, &col, &bytePos);
+
+    if (strcmp(filename, "test.xml") != 0 ||
+        line != 4 || col != 11 || bytePos != 62) {
+        fprintf(stderr, "unexpected position: %s %d %d %lu\n",
+                filename, line, col, bytePos);
+        *err = 1;
+    }
+
+    size = 80;
+    xmlCtxtGetInputWindow(ctxt, 0, &start, &size, &offset);
+
+    if (strncmp((char *) start, "<doc>&ent;", 10) != 0 ||
+        size != 16 || offset != 10) {
+        fprintf(stderr, "unexpected window: %.10s %d %d\n",
+                start, size, offset);
+        *err = 1;
+    }
+
+    xmlCtxtGetInputPosition(ctxt, -1, &filename, &line, &col, &bytePos);
+
+    if (filename != NULL ||
+        line != 1 || col != 11 || bytePos != 10) {
+        fprintf(stderr, "unexpected position: %s %d %d %lu\n",
+                filename, line, col, bytePos);
+        *err = 1;
+    }
+
+    size = 80;
+    xmlCtxtGetInputWindow(ctxt, -1, &start, &size, &offset);
+
+    if (strncmp((char *) start, "xxx &fail;", 10) != 0 ||
+        size != 14 || offset != 10) {
+        fprintf(stderr, "unexpected window: %.10s %d %d\n",
+                start, size, offset);
+        *err = 1;
+    }
+}
+
+static int
+testCtxtInputGetters(void) {
+    const char *xml =
+        "<!DOCTYPE doc [\n"
+        "  <!ENTITY ent 'xxx &fail; xxx'>\n"
+        "]>\n"
+        "<doc>&ent;</doc>\n";
+    xmlParserCtxt *ctxt;
+    xmlDoc *doc;
+    int err = 0;
+
+    ctxt = xmlNewParserCtxt();
+    xmlCtxtSetErrorHandler(ctxt, testCtxtInputGetterError, &err);
+    doc = xmlCtxtReadDoc(ctxt, BAD_CAST xml, "test.xml", NULL, 0);
+    xmlFreeDoc(doc);
+    xmlFreeParserCtxt(ctxt);
+
+    return err;
+}
+
 #ifdef LIBXML_VALID_ENABLED
 static void
 testSwitchDtdExtSubset(void *vctxt, const xmlChar *name ATTRIBUTE_UNUSED,
@@ -1428,6 +1497,7 @@ main(void) {
     err |= testCFileIO();
     err |= testUndeclEntInContent();
     err |= testInvalidCharRecovery();
+    err |= testCtxtInputGetters();
 #ifdef LIBXML_VALID_ENABLED
     err |= testSwitchDtd();
 #endif