start to alter parseing to read from cos_stream object

10 files changed, 212 insertions, 74 deletions

diff --git a/include/nspdf/document.h b/include/nspdf/document.h
index d7cbb0f..3c222cf 100644
--- a/include/nspdf/document.h
+++ b/include/nspdf/document.h
@@ -40,7 +40,7 @@ nspdferror nspdf_document_destroy(struct nspdf_doc *doc);
  * ready to render pages. The passed buffer ownership is transfered and must
  * not be altered untill the document is destroyed.
  */
-nspdferror nspdf_document_parse(struct nspdf_doc *doc, const uint8_t *buffer, uint64_t buffer_length);
+nspdferror nspdf_document_parse(struct nspdf_doc *doc, const uint8_t *buffer, unsigned int buffer_length);
 
 
 #endif /* NSPDF_DOCUMENT_H_ */
diff --git a/src/cos_object.c b/src/cos_object.c
index 3dc5efa..4398822 100644
--- a/src/cos_object.c
+++ b/src/cos_object.c
@@ -17,6 +17,7 @@
 #include <nspdf/errors.h>
 
 #include "cos_object.h"
+#include "cos_parse.h"
 #include "pdf_doc.h"
 
 
@@ -369,6 +370,54 @@ cos_get_stream(struct nspdf_doc *doc,
 
 
 /*
+ * get object from object reference
+ */
+nspdferror
+cos_get_object(struct nspdf_doc *doc,
+               struct cos_object *cobj,
+               struct cos_object **value_out)
+{
+    nspdferror res;
+    res = nspdf__xref_get_referenced(doc, &cobj);
+    if (res == NSPDFERROR_OK) {
+        *value_out = cobj;
+    }
+    return res;
+}
+
+
+nspdferror
+cos_get_content(struct nspdf_doc *doc,
+                struct cos_object *cobj,
+                struct cos_content **content_out)
+{
+    nspdferror res;
+    struct cos_object *content_obj;
+
+    res = nspdf__xref_get_referenced(doc, &cobj);
+    if (res == NSPDFERROR_OK) {
+        if (cobj->type == COS_TYPE_STREAM) {
+            res = cos_parse_content_stream(doc, cobj->u.stream, &content_obj);
+            if (res == NSPDFERROR_OK) {
+                /* replace stream object with parsed content operations */
+                struct cos_object tmpobj;
+                tmpobj = *cobj;
+                *cobj = *content_obj;
+                *content_obj = tmpobj;
+                cos_free_object(content_obj);
+
+                *content_out = cobj->u.content;
+            }
+        } else if (cobj->type == COS_TYPE_CONTENT) {
+            *content_out = cobj->u.content;
+        } else {
+            res = NSPDFERROR_TYPE;
+        }
+    }
+    return res;
+}
+
+/*
  * get a value for a key from a dictionary
  */
 nspdferror
diff --git a/src/cos_object.h b/src/cos_object.h
index 2e763e2..9b98694 100644
--- a/src/cos_object.h
+++ b/src/cos_object.h
@@ -16,6 +16,7 @@
 #define NSPDF__COS_OBJECT_H_
 
 struct nspdf_doc;
+struct content_operation;
 
 enum cos_type {
     COS_TYPE_NULL, /* 0 */
@@ -30,6 +31,7 @@ enum cos_type {
     COS_TYPE_NUMBERTREE,
     COS_TYPE_STREAM,
     COS_TYPE_REFERENCE, /* 11 */
+    COS_TYPE_CONTENT, /* 12 - parsed content stream */
 };
 
 struct cos_object;
@@ -59,10 +61,13 @@ struct cos_array {
     struct cos_object **values;
 };
 
+/**
+ * COS string data
+ */
 struct cos_string {
-    uint8_t *data; /**< string data */
-    size_t length; /**< string length */
+    unsigned int length; /**< string length */
     size_t alloc; /**< memory allocation for string */
+    uint8_t *data; /**< string data */
 };
 
 struct cos_reference {
@@ -71,12 +76,22 @@ struct cos_reference {
 };
 
 struct cos_stream {
-    const uint8_t *data; /**< decoded stream data */
-    int64_t length; /**< decoded stream length */
+    unsigned int length; /**< decoded stream length */
     size_t alloc; /**< memory allocated for stream */
+    const uint8_t *data; /**< decoded stream data */
 };
 
 
+/**
+ * Synthetic parsed content object.
+ *
+ */
+struct cos_content {
+    unsigned int length; /**< number of content operations */
+    unsigned int alloc; /**< number of allocated operations */
+    struct content_operation *operations;
+};
+
 struct cos_object {
     int type;
     union {
@@ -107,6 +122,8 @@ struct cos_object {
         /** reference */
         struct cos_reference *reference;
 
+        /** parsed content stream */
+        struct cos_content *content;
     } u;
 };
 
@@ -264,5 +281,31 @@ nspdferror cos_get_array(struct nspdf_doc *doc, struct cos_object *cobj, struct
  */
 nspdferror cos_get_stream(struct nspdf_doc *doc, struct cos_object *cobj, struct cos_stream **stream_out);
 
+/**
+ * get a direct cos object.
+ *
+ * Obtain a direct object if the passed object was a reference it is
+ * dereferenced from the cross reference table.
+ *
+ * \param doc The document the cos object belongs to.
+ * \param cobj A cos object.
+ * \param object_out The result object.
+ * \return NSERROR_OK and \p object_out updated,
+ */
+nspdferror cos_get_object(struct nspdf_doc *doc, struct cos_object *cobj, struct cos_object **object_out);
+
+/**
+ * get a parsed content object
+ *
+ * Get the parsed content from a cos object, if the object is an object
+ *   reference it will be dereferenced first.
+ * The parsed content object is *not* a normal COS object rather it is the
+ *   internal result of parsing a PDF content stream.
+ * This object type is used to replace the stream object in the cross reference
+ *   table after its initial parse to avoid the need to keep and repeatedly
+ *   parse the filtered stream data.
+ *
+ */
+nspdferror cos_get_content(struct nspdf_doc *doc, struct cos_object *cobj, struct cos_content **content_out);
 
 #endif
diff --git a/src/cos_parse.c b/src/cos_parse.c
index 21ba0d7..c196019 100644
--- a/src/cos_parse.c
+++ b/src/cos_parse.c
@@ -57,11 +57,15 @@ static uint8_t xtoi(uint8_t x)
     return x;
 }
 
+/**
+ * parse a number
+ */
 static nspdferror
-cos_decode_number(struct nspdf_doc *doc,
-                      uint64_t *offset_out,
-                      struct cos_object **cosobj_out)
+cos_parse_number(struct cos_stream *stream,
+                  uint64_t *offset_out,
+                  struct cos_object **cosobj_out)
 {
+    nspdferror res;
     struct cos_object *cosobj;
     uint8_t c; /* current byte from source data */
     unsigned int len; /* number of decimal places in number */
@@ -71,7 +75,7 @@ cos_decode_number(struct nspdf_doc *doc,
     offset = *offset_out;
 
     for (len = 0; len < sizeof(num); len++) {
-        c = DOC_BYTE(doc, offset);
+        c = stream_byte(stream, offset);
         if ((bclass[c] & BC_DCML) != BC_DCML) {
             int64_t result = 0; /* parsed result */
             uint64_t tens;
@@ -85,7 +89,10 @@ cos_decode_number(struct nspdf_doc *doc,
                 result += (num[len - 1] * tens);
             }
 
-            doc_skip_ws(doc, &offset);
+            res = nspdf__stream_skip_ws(stream, &offset);
+            if (res != NSPDFERROR_OK) {
+                return res;
+            }
 
             cosobj = calloc(1, sizeof(struct cos_object));
             if (cosobj == NULL) {
@@ -230,7 +237,7 @@ cos_decode_string(struct nspdf_doc *doc,
         cos_string_append(cstring, c);
     }
 
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     *cosobj_out = cosobj;
     *offset_out = offset;
@@ -279,7 +286,7 @@ cos_decode_hex_string(struct nspdf_doc *doc,
                 cos_string_append(cstring, value);
             }
             offset++;
-            doc_skip_ws(doc, &offset);
+            nspdf__stream_skip_ws(doc->stream, &offset);
 
             *cosobj_out = cosobj;
             *offset_out = offset;
@@ -323,7 +330,7 @@ cos_decode_dictionary(struct nspdf_doc *doc,
         return NSPDFERROR_SYNTAX; /* syntax error */
     }
     offset += 2;
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     //printf("found a dictionary\n");
 
@@ -371,7 +378,7 @@ cos_decode_dictionary(struct nspdf_doc *doc,
 
     }
     offset += 2; /* skip closing >> */
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     *cosobj_out = cosobj;
     *offset_out = offset;
@@ -403,7 +410,7 @@ cos_parse_list(struct nspdf_doc *doc,
     offset++;
 
     /* advance offset to next token */
-    res = doc_skip_ws(doc, &offset);
+    res = nspdf__stream_skip_ws(doc->stream, &offset);
     if (res != NSPDFERROR_OK) {
         return res;
     }
@@ -449,7 +456,7 @@ cos_parse_list(struct nspdf_doc *doc,
     }
     offset++; /* skip closing ] */
 
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     *cosobj_out = cosobj;
     *offset_out = offset;
@@ -499,7 +506,7 @@ cos_decode_name(struct nspdf_doc *doc,
 
     //printf("name: %s\n", name);
 
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     cosobj = calloc(1, sizeof(struct cos_object));
     if (cosobj == NULL) {
@@ -575,7 +582,7 @@ cos_decode_boolean(struct nspdf_doc *doc,
         return -1; /* syntax error */
     }
 
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     cosobj = calloc(1, sizeof(struct cos_object));
     if (cosobj == NULL) {
@@ -626,7 +633,7 @@ cos_decode_null(struct nspdf_doc *doc,
         return -1; /* syntax error */
     }
 
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     cosobj = calloc(1, sizeof(struct cos_object));
     if (cosobj == NULL) {
@@ -656,6 +663,7 @@ cos_parse_stream(struct nspdf_doc *doc,
     uint64_t offset;
     struct cos_object *stream_filter;
     struct cos_stream *stream;
+    int64_t stream_length;
 
     offset = *offset_out;
     stream_dict = *cosobj_out;
@@ -678,7 +686,7 @@ cos_parse_stream(struct nspdf_doc *doc,
     //printf("detected stream\n");
 
     /* parsed object was a dictionary and there is a stream marker */
-    res = doc_skip_ws(doc, &offset);
+    res = nspdf__stream_skip_ws(doc->stream, &offset);
     if (res != NSPDFERROR_OK) {
         return res;
     }
@@ -688,10 +696,15 @@ cos_parse_stream(struct nspdf_doc *doc,
         return NSPDFERROR_NOMEM;
     }
 
-    res = cos_get_dictionary_int(doc, stream_dict, "Length", &stream->length);
+    res = cos_get_dictionary_int(doc, stream_dict, "Length", &stream_length);
     if (res != NSPDFERROR_OK) {
         return res;
     }
+    if (stream_length < 0) {
+        return NSPDFERROR_RANGE;
+    }
+    stream->length = stream_length;
+
     //printf("stream length %d\n", stream_length);
     stream->data = doc->start + offset;
     stream->alloc = 0; /* stream is pointing at non malloced data */
@@ -699,7 +712,7 @@ cos_parse_stream(struct nspdf_doc *doc,
     offset += stream->length;
 
     /* possible whitespace after stream data */
-    res = doc_skip_ws(doc, &offset);
+    res = nspdf__stream_skip_ws(doc->stream, &offset);
     if (res != NSPDFERROR_OK) {
         return res;
     }
@@ -719,7 +732,7 @@ cos_parse_stream(struct nspdf_doc *doc,
     offset += 9;
     //printf("detected endstream\n");
 
-    res = doc_skip_ws(doc, &offset);
+    res = nspdf__stream_skip_ws(doc->stream, &offset);
     if (res != NSPDFERROR_OK) {
         return res;
     }
@@ -785,7 +798,7 @@ cos_attempt_decode_reference(struct nspdf_doc *doc,
 
     offset = *offset_out;
 
-    res = cos_decode_number(doc, &offset, &generation);
+    res = cos_parse_number(doc->stream, &offset, &generation);
     if (res != NSPDFERROR_OK) {
         /* no error if next token could not be decoded as a number */
         return NSPDFERROR_OK;
@@ -813,7 +826,7 @@ cos_attempt_decode_reference(struct nspdf_doc *doc,
         //printf("found object reference\n");
         offset ++;
 
-        doc_skip_ws(doc, &offset);
+        nspdf__stream_skip_ws(doc->stream, &offset);
 
         nref = calloc(1, sizeof(struct cos_reference));
         if (nref == NULL) {
@@ -839,7 +852,7 @@ cos_attempt_decode_reference(struct nspdf_doc *doc,
         //printf("indirect\n");
         offset += 3;
 
-        res = doc_skip_ws(doc, &offset);
+        res = nspdf__stream_skip_ws(doc->stream, &offset);
         if (res != NSPDFERROR_OK) {
             cos_free_object(generation);
             return res;
@@ -880,7 +893,7 @@ cos_attempt_decode_reference(struct nspdf_doc *doc,
         offset += 6;
         //printf("endobj\n");
 
-        res = doc_skip_ws(doc, &offset);
+        res = nspdf__stream_skip_ws(doc->stream, &offset);
         if (res != NSPDFERROR_OK) {
             cos_free_object(indirect);
             cos_free_object(generation);
@@ -980,7 +993,7 @@ cos_parse_object(struct nspdf_doc *doc,
     case '7':
     case '8':
     case '9':
-        res = cos_decode_number(doc, &offset, &cosobj);
+        res = cos_parse_number(doc->stream, &offset, &cosobj);
         /* if type is positive integer try to check for reference */
         if ((res == 0) &&
             (cosobj->type == COS_TYPE_INT) &&
@@ -1032,3 +1045,23 @@ cos_parse_object(struct nspdf_doc *doc,
 
     return res;
 }
+
+nspdferror
+cos_parse_content_stream(struct nspdf_doc *doc,
+                         struct cos_stream *stream,
+                         struct cos_object **content_out)
+{
+    struct cos_object *cosobj;
+
+    printf("%.*s", (int)stream->length, stream->data);
+
+    cosobj = calloc(1, sizeof(struct cos_object));
+    if (cosobj == NULL) {
+        return NSPDFERROR_NOMEM;
+    }
+    cosobj->type = COS_TYPE_CONTENT;
+
+    *content_out = cosobj;
+
+    return NSPDFERROR_OK;
+}
diff --git a/src/cos_parse.h b/src/cos_parse.h
index 0bca79f..8f48108 100644
--- a/src/cos_parse.h
+++ b/src/cos_parse.h
@@ -17,12 +17,18 @@
 
 struct nspdf_doc;
 struct cos_object;
+struct cos_stream;
 
 /**
- * Decode input stream into an object
+ * Parse input stream into an object
  *
  * lex and parse a byte stream to generate a COS object.
  */
 nspdferror cos_parse_object(struct nspdf_doc *doc, uint64_t *offset_out, struct cos_object **cosobj_out);
 
+/**
+ * Parse content stream into content operations object
+ */
+nspdferror cos_parse_content_stream(struct nspdf_doc *doc, struct cos_stream *stream, struct cos_object **content_out);
+
 #endif
diff --git a/src/document.c b/src/document.c
index bbe948d..b7a36d2 100644
--- a/src/document.c
+++ b/src/document.c
@@ -92,7 +92,7 @@ decode_startxref(struct nspdf_doc *doc,
     }
     offset += 9;
 
-    res = doc_skip_ws(doc, &offset);
+    res = nspdf__stream_skip_ws(doc->stream, &offset);
     if (res != NSPDFERROR_OK) {
         return res;
     }
@@ -168,7 +168,7 @@ decode_trailer(struct nspdf_doc *doc,
         return -1;
     }
     offset += 7;
-    doc_skip_ws(doc, &offset);
+    nspdf__stream_skip_ws(doc->stream, &offset);
 
     res = cos_parse_object(doc, &offset, &trailer);
     if (res != 0) {
@@ -422,13 +422,20 @@ static nspdferror check_header(struct nspdf_doc *doc)
 nspdferror
 nspdf_document_parse(struct nspdf_doc *doc,
                      const uint8_t *buffer,
-                     uint64_t buffer_length)
+                     unsigned int buffer_length)
 {
     nspdferror res;
 
     doc->start = buffer;
     doc->length = buffer_length;
 
+    doc->stream = calloc(1, sizeof(struct cos_stream));
+    if (doc->stream == NULL) {
+        return NSPDFERROR_NOMEM;
+    }
+    doc->stream->data = buffer;
+    doc->stream->length = buffer_length;
+
     res = check_header(doc);
     if (res != 0) {
         printf("header check failed\n");
diff --git a/src/page.c b/src/page.c
index acc97d7..7b6bee8 100644
--- a/src/page.c
+++ b/src/page.c
@@ -148,10 +148,18 @@ nspdf_page_count(struct nspdf_doc *doc, unsigned int *pages_out)
 
 static nspdferror
 nspdf__render_content_stream(struct nspdf_doc *doc,
-                             struct cos_stream *content_stream)
+                             struct page_table_entry *page_entry,
+                             struct cos_object *content_entry)
 {
-    printf("%.*s", (int)content_stream->length, content_stream->data);
-    return NSPDFERROR_OK;
+    nspdferror res;
+    struct cos_content *content_operations;
+
+    res = cos_get_content(doc, content_entry, &content_operations);
+    if (res == NSPDFERROR_OK) {
+        printf("%p", content_operations);
+    }
+
+    return res;
 }
 
 /* exported interface documented in nspdf/page.h */
@@ -160,7 +168,6 @@ nspdf_page_render(struct nspdf_doc *doc, unsigned int page_number)
 {
     struct page_table_entry *page_entry;
     struct cos_object *content_array;
-    struct cos_stream *content_stream;
     nspdferror res;
 
     page_entry = doc->page_table + page_number;
@@ -184,31 +191,16 @@ nspdf_page_render(struct nspdf_doc *doc, unsigned int page_number)
                                 content_stream_index,
                                 &content_entry);
             if (res != NSPDFERROR_OK) {
-                return res;
-            }
-
-            res = cos_get_stream(doc, content_entry, &content_stream);
-            if (res != NSPDFERROR_OK) {
-                return res;
+                break;
             }
 
-            res = nspdf__render_content_stream(doc, content_stream);
+            res = nspdf__render_content_stream(doc, page_entry, content_entry);
             if (res != NSPDFERROR_OK) {
-                return res;
+                break;
             }
         }
     } else if (res == NSPDFERROR_TYPE) {
-        res = cos_get_stream(doc, page_entry->contents, &content_stream);
-        if (res != NSPDFERROR_OK) {
-            return res;
-        }
-
-        res = nspdf__render_content_stream(doc, content_stream);
-        if (res != NSPDFERROR_OK) {
-            return res;
-        }
-    } else {
-        return res;
+        res = nspdf__render_content_stream(doc, page_entry, page_entry->contents);
     }
 
     return res;
diff --git a/src/pdf_doc.c b/src/pdf_doc.c
index 997a3d7..955f737 100644
--- a/src/pdf_doc.c
+++ b/src/pdf_doc.c
@@ -19,29 +19,27 @@
 #include "cos_object.h"
 #include "pdf_doc.h"
 
-/**
- * move offset to next non whitespace byte
- */
-nspdferror doc_skip_ws(struct nspdf_doc *doc, uint64_t *offset)
+nspdferror nspdf__stream_skip_ws(struct cos_stream *stream, uint64_t *offset)
 {
     uint8_t c;
     /* TODO sort out keeping offset in range */
-    c = DOC_BYTE(doc, *offset);
+    c = stream_byte(stream, *offset);
     while ((bclass[c] & (BC_WSPC | BC_CMNT) ) != 0) {
         (*offset)++;
         /* skip comments */
         if ((bclass[c] & BC_CMNT) != 0) {
-            c = DOC_BYTE(doc, *offset);
+            c = stream_byte(stream, *offset);
             while ((bclass[c] & BC_EOLM ) == 0) {
                 (*offset)++;
-                c = DOC_BYTE(doc, *offset);
+                c = stream_byte(stream, *offset);
             }
         }
-        c = DOC_BYTE(doc, *offset);
+        c = stream_byte(stream, *offset);
     }
     return NSPDFERROR_OK;
 }
 
+
 /**
  * move offset to next non eol byte
  */
diff --git a/src/pdf_doc.h b/src/pdf_doc.h
index e362ea6..27a730a 100644
--- a/src/pdf_doc.h
+++ b/src/pdf_doc.h
@@ -18,11 +18,18 @@
 struct xref_table_entry;
 struct page_table_entry;
 
-/** pdf document */
+/**
+ * pdf document
+ */
 struct nspdf_doc {
 
     const uint8_t *start; /* start of pdf document in input stream */
-    uint64_t length;
+    unsigned int length;
+
+    /**
+     * input data stream
+     */
+    struct cos_stream *stream;
 
     int major;
     int minor;
@@ -46,8 +53,14 @@ struct nspdf_doc {
 /* byte data acessory, allows for more complex buffer handling in future */
 #define DOC_BYTE(doc, offset) (doc->start[(offset)])
 
+static inline uint8_t
+stream_byte(struct cos_stream *stream, unsigned int offset)
+{
+    return *(stream->data + offset);
+}
+
 /* helpers in pdf_doc.c */
-nspdferror doc_skip_ws(struct nspdf_doc *doc, uint64_t *offset);
+nspdferror nspdf__stream_skip_ws(struct cos_stream *stream, uint64_t *offset);
 nspdferror doc_skip_eol(struct nspdf_doc *doc, uint64_t *offset);
 nspdferror doc_read_uint(struct nspdf_doc *doc, uint64_t *offset_out, uint64_t *result_out);
 
@@ -68,9 +81,6 @@ nspdferror nspdf__xref_allocate(struct nspdf_doc *doc, int64_t size);
 nspdferror nspdf__decode_page_tree(struct nspdf_doc *doc, struct cos_object *page_tree_node, unsigned int *page_index);
 
 /* cos stream filters */
-nspdferror
-nspdf__cos_stream_filter(struct nspdf_doc *doc,
-                         const char *filter_name,
-                         struct cos_stream **stream_out);
+nspdferror nspdf__cos_stream_filter(struct nspdf_doc *doc, const char *filter_name, struct cos_stream **stream_out);
 
 #endif
diff --git a/src/xref.c b/src/xref.c
index 298c750..2fb9301 100644
--- a/src/xref.c
+++ b/src/xref.c
@@ -68,7 +68,7 @@ nspdferror nspdf__xref_parse(struct nspdf_doc *doc, uint64_t *offset_out)
     }
     offset += 4;
 
-    res = doc_skip_ws(doc, &offset);
+    res = nspdf__stream_skip_ws(doc->stream, &offset);
     if (res != NSPDFERROR_OK) {
         return res;
     }
@@ -79,7 +79,7 @@ nspdferror nspdf__xref_parse(struct nspdf_doc *doc, uint64_t *offset_out)
     res = doc_read_uint(doc, &offset, &objnumber);
     while (res == NSPDFERROR_OK) {
         uint64_t lastobj;
-        res = doc_skip_ws(doc, &offset);
+        res = nspdf__stream_skip_ws(doc->stream, &offset);
         if (res != NSPDFERROR_OK) {
             return res;
         }
@@ -89,7 +89,7 @@ nspdferror nspdf__xref_parse(struct nspdf_doc *doc, uint64_t *offset_out)
             return res;
         }
 
-        res = doc_skip_ws(doc, &offset);
+        res = nspdf__stream_skip_ws(doc->stream, &offset);
         if (res != NSPDFERROR_OK) {
             return res;
         }
@@ -169,7 +169,7 @@ nspdf__xref_get_referenced(struct nspdf_doc *doc, struct cos_object **cobj_out)
     }
 
     if (entry->object == NULL) {
-        /* indirect object has never been decoded */
+        /* indirect object has never been parsed */
         offset = entry->offset;
         res = cos_parse_object(doc, &offset, &indirect);
         if (res != NSPDFERROR_OK) {