Add SAM header iterator methods [DRAFT]

Author: jmarshall

header.c

@@ -1271,6 +1271,18 @@ int sam_hdr_rebuild(sam_hdr_t *bh) {
     return 0;
 }
 
+sam_hdr_line_t *sam_hdr_first_line(sam_hdr_t *bh) {
+    if (!bh->hrecs) {
+        if (sam_hdr_fill_hrecs(bh) != 0) return NULL;
+    }
+    return bh->hrecs->first_line;
+}
+
+sam_hdr_line_t *sam_hdr_next_line(sam_hdr_t *bh, sam_hdr_line_t *line) {
+    line = line->global_next;
+    return (line != bh->hrecs->first_line)? line : NULL;
+}
+
 /*
  * Appends a formatted line to an existing SAM header.
  * Line is a full SAM header record, eg "@SQ\tSN:foo\tLN:100", with
@@ -1346,6 +1358,13 @@ int sam_hdr_add_line(sam_hdr_t *bh, const char *type, ...) {
     return ret;
 }
 
+int sam_hdr_format_line_append(const sam_hdr_line_t *line, kstring_t *ks) {
+    if (!line) return -1;
+
+    if (build_header_line(line, ks) < 0) return -2;
+    return 0;
+}
+
 /*
  * Returns a complete line of formatted text for a specific head type/ID
  * combination. If ID_key is NULL then it returns the first line of the specified
@@ -1399,6 +1418,27 @@ int sam_hdr_find_line_pos(sam_hdr_t *bh, const char *type,
     return 0;
 }
 
+/*
+ * Remove a line from the header via an iterator.
+ */
+
+sam_hdr_line_t *sam_hdr_remove_line(sam_hdr_t *bh, sam_hdr_line_t *line) {
+    if (!bh || !line) return NULL;
+
+    if (line->type == TYPEKEY("PG")) {
+        hts_log_warning("Removing PG lines is not supported!");
+        return NULL;
+    }
+
+    sam_hdr_line_t *next = sam_hdr_next_line(bh, line);
+    char type[2] = { line->type >> 8, line->type & 0xff };
+    if (sam_hrecs_remove_line(bh->hrecs, type, line, 1) < 0) return NULL;
+
+    if (bh->hrecs->refs_changed >= 0 && rebuild_target_arrays(bh) != 0) return NULL;
+    if (bh->hrecs->dirty) redact_header_text(bh);
+    return next;
+}
+
 /*
  * Remove a line from the header by specifying a tag:value that uniquely
  * identifies a line, i.e. the @SQ line containing "SN:ref1".
@@ -1908,6 +1948,19 @@ const char *sam_hdr_line_name(sam_hdr_t *bh,
 
 /* ==== Key:val level methods ==== */
 
+int sam_hdr_find_tag(const sam_hdr_line_t *line,
+                     const char *key,
+                     kstring_t *ks) {
+    if (!line || !key) return -2;
+
+    sam_hrec_tag_t *tag = sam_hrecs_find_key(line, key, NULL);
+    if (!tag || tag->len < 3) return -1;
+
+    ks_clear(ks);
+    if (kputsn(&tag->str[3], tag->len-3, ks) < 0) return -2;
+    return 0;
+}
+
 int sam_hdr_find_tag_id(sam_hdr_t *bh,
                      const char *type,
                      const char *ID_key,
@@ -1940,6 +1993,16 @@ int sam_hdr_find_tag_id(sam_hdr_t *bh,
     return 0;
 }
 
+int sam_hdr_remove_tag(sam_hdr_t *bh,
+                       sam_hdr_line_t *line,
+                       const char *key) {
+    if (!bh || !line || !key) return -1;
+
+    int ret = sam_hrecs_remove_key(bh->hrecs, line, key);
+    if (ret == 0 && bh->hrecs->dirty) redact_header_text(bh);
+    return ret;
+}
+
 int sam_hdr_find_tag_pos(sam_hdr_t *bh,
                      const char *type,
                      int pos,
@@ -2635,7 +2698,7 @@ static int sam_hrecs_update(sam_hrecs_t *hrecs, sam_hrec_type_t *type, ...) {
  * Returns the tag pointer on success
  *         NULL on failure
  */
-sam_hrec_tag_t *sam_hrecs_find_key(sam_hrec_type_t *type,
+sam_hrec_tag_t *sam_hrecs_find_key(const sam_hrec_type_t *type,
                                    const char *key,
                                    sam_hrec_tag_t **prev) {
     sam_hrec_tag_t *tag, *p = NULL;

header.h

@@ -290,7 +290,7 @@ void sam_hrecs_free(sam_hrecs_t *hrecs);
 sam_hrec_type_t *sam_hrecs_find_type_id(sam_hrecs_t *hrecs, const char *type,
                                      const char *ID_key, const char *ID_value);
 
-sam_hrec_tag_t *sam_hrecs_find_key(sam_hrec_type_t *type,
+sam_hrec_tag_t *sam_hrecs_find_key(const sam_hrec_type_t *type,
                                    const char *key,
                                    sam_hrec_tag_t **prev);
 

htslib/sam.h

@@ -44,6 +44,8 @@ DEALINGS IN THE SOFTWARE.  */
 extern "C" {
 #endif
 
+struct sam_hrec_type_s;
+
 /// Highest SAM format version supported by this library
 #define SAM_FORMAT_VERSION "1.6"
 
@@ -466,6 +468,22 @@ int sam_hdr_nref(const sam_hdr_t *h);
 
 /* ==== Line level methods ==== */
 
+/*! @typedef
+ * @abstract Opaque type used as an iterator over header lines.
+ */
+typedef struct sam_hrec_type_s sam_hdr_line_t;
+
+/// Return an iterator pointing to the first header line
+HTSLIB_EXPORT
+sam_hdr_line_t *sam_hdr_first_line(sam_hdr_t *h);
+
+/// Return an iterator pointing to the next header line
+/*!
+ * @return  An iterator pointing to the next line, or NULL if there is none.
+ */
+HTSLIB_EXPORT
+sam_hdr_line_t *sam_hdr_next_line(sam_hdr_t *h, sam_hdr_line_t *line);
+
 /// Add formatted lines to an existing header.
 /*!
  * @param lines  Full SAM header record, eg "@SQ\tSN:foo\tLN:100", with
@@ -498,6 +516,36 @@ int sam_hdr_add_lines(sam_hdr_t *h, const char *lines, size_t len);
 HTSLIB_EXPORT
 int sam_hdr_add_line(sam_hdr_t *h, const char *type, ...);
 
+/// Returns a complete line of formatted text for the line pointed to.
+/*!
+ * @param line      Iterator pointing to a header line
+ * @param ks        kstring to which to append the result
+ * @return          0 on success;
+ *                 -1 if @p line does not point to a header line
+ *                 -2 on other failures
+ *
+ * Puts a complete line of formatted text for a specific line into @p ks.
+ * Appends the text to the existing content in @p ks, if any.
+ */
+HTSLIB_EXPORT
+int sam_hdr_format_line_append(const sam_hdr_line_t *line, kstring_t *ks);
+
+/// Returns a complete line of formatted text for the line pointed to.
+/*!
+ * @param line      Iterator pointing to a header line
+ * @param ks        kstring to hold the result
+ * @return          0 on success;
+ *                 -1 if @p line does not point to a header line
+ *                 -2 on other failures
+ *
+ * Puts a complete line of formatted text for a specific line into @p ks.
+ * Any existing content in @p ks will be overwritten.
+ */
+static inline int sam_hdr_format_line(const sam_hdr_line_t *line, kstring_t *ks)
+{
+    return sam_hdr_format_line_append(line, ks_clear(ks));
+}
+
 /// Returns a complete line of formatted text for a given type and ID.
 /*!
  * @param type      Type of the searched line. Eg. "SQ"
@@ -536,6 +584,14 @@ HTSLIB_EXPORT
 int sam_hdr_find_line_pos(sam_hdr_t *h, const char *type,
                           int pos, kstring_t *ks);
 
+/// Remove line pointed to by iterator from a header
+/*!
+ * @param line     Iterator pointing to a header line
+ * @return         An iterator pointing to the following line, or NULL on error FIXME or if it was the last line
+ */
+HTSLIB_EXPORT
+sam_hdr_line_t *sam_hdr_remove_line(sam_hdr_t *h, sam_hdr_line_t *line);
+
 /// Remove a line with given type / id from a header
 /*!
  * @param type      Type of the searched line. Eg. "SQ"
@@ -687,6 +743,21 @@ const char *sam_hdr_line_name(sam_hdr_t *bh, const char *type, int pos);
 
 /* ==== Key:val level methods ==== */
 
+/// Return the value associated with a key for a header line identified by iterator
+/*!
+ * @param line      Iterator pointing to a header line
+ * @param key       Key of the searched tag. Eg. "LN"
+ * @param ks        kstring where the value will be written
+ * @return          0 on success
+ *                 -1 if the requested tag does not exist
+ *                 -2 on other errors
+ *
+ * Looks for a specific key in the SAM header line pointed to by @p line and writes the
+ * associated value into @p ks.  Any pre-existing content in @p ks will be overwritten.
+ */
+HTSLIB_EXPORT
+int sam_hdr_find_tag(const sam_hdr_line_t *line, const char *key, kstring_t *ks);
+
 /// Return the value associated with a key for a header line identified by ID_key:ID_val
 /*!
  * @param type      Type of the line to which the tag belongs. Eg. "SQ"
@@ -724,6 +795,15 @@ int sam_hdr_find_tag_id(sam_hdr_t *h, const char *type, const char *ID_key, cons
 HTSLIB_EXPORT
 int sam_hdr_find_tag_pos(sam_hdr_t *h, const char *type, int pos, const char *key, kstring_t *ks);
 
+/// Remove the key from the line pointed to by the iterator.
+/*!
+ * @param line      Iterator pointing to a header line
+ * @param key       Key of the targeted tag. Eg. "M5"
+ * @return          1 if the key was removed; 0 if it was not present; -1 on error
+ */
+HTSLIB_EXPORT
+int sam_hdr_remove_tag(sam_hdr_t *h, sam_hdr_line_t *line, const char *key);
+
 /// Remove the key from the line identified by type, ID_key and ID_value.
 /*!
  * @param type      Type of the line to which the tag belongs. Eg. "SQ"