dovecot-2.1: fts-solr: Fixed indexing messages with multiple MIM...

[dovecot at dovecot.org]

details:   http://hg.dovecot.org/dovecot-2.1/rev/283e08771d92
changeset: 13435:283e08771d92
user:      Timo Sirainen <tss at iki.fi>
date:      Thu Sep 08 16:37:56 2011 +0300
description:
fts-solr: Fixed indexing messages with multiple MIME body parts.

diffstat:

 src/plugins/fts-solr/fts-backend-solr-old.c |  18 ++++++++++++++----
 src/plugins/fts-solr/fts-backend-solr.c     |  19 +++++++++++++++----
 2 files changed, 29 insertions(+), 8 deletions(-)

diffs (103 lines):

diff -r 1aa51cd11614 -r 283e08771d92 src/plugins/fts-solr/fts-backend-solr-old.c
--- a/src/plugins/fts-solr/fts-backend-solr-old.c	Thu Sep 08 12:39:13 2011 +0300
+++ b/src/plugins/fts-solr/fts-backend-solr-old.c	Thu Sep 08 16:37:56 2011 +0300
@@ -36,6 +36,7 @@
 	string_t *cmd, *hdr;
 
 	bool headers_open;
+	bool body_open;
 	bool documents_added;
 };
 
@@ -513,6 +514,10 @@
 		str_append(ctx->cmd, "<add>");
 	} else {
 		ctx->headers_open = FALSE;
+		if (ctx->body_open) {
+			ctx->body_open = FALSE;
+			str_append(ctx->cmd, "</field>");
+		}
 		str_append(ctx->cmd, "<field name=\"hdr\">");
 		str_append_str(ctx->cmd, ctx->hdr);
 		str_append(ctx->cmd, "</field>");
@@ -547,7 +552,10 @@
 		break;
 	case FTS_BACKEND_BUILD_KEY_BODY_PART:
 		ctx->headers_open = FALSE;
-		str_append(ctx->cmd, "<field name=\"body\">");
+		if (!ctx->body_open) {
+			ctx->body_open = TRUE;
+			str_append(ctx->cmd, "<field name=\"body\">");
+		}
 		break;
 	case FTS_BACKEND_BUILD_KEY_BODY_PART_BINARY:
 		i_unreached();
@@ -561,10 +569,12 @@
 	struct solr_fts_backend_update_context *ctx =
 		(struct solr_fts_backend_update_context *)_ctx;
 
-	if (!ctx->headers_open)
-		str_append(ctx->cmd, "</field>");
-	else
+	if (ctx->headers_open)
 		str_append_c(ctx->hdr, '\n');
+	else {
+		i_assert(ctx->body_open);
+		str_append_c(ctx->cmd, '\n');
+	}
 }
 
 static int
diff -r 1aa51cd11614 -r 283e08771d92 src/plugins/fts-solr/fts-backend-solr.c
--- a/src/plugins/fts-solr/fts-backend-solr.c	Thu Sep 08 12:39:13 2011 +0300
+++ b/src/plugins/fts-solr/fts-backend-solr.c	Thu Sep 08 16:37:56 2011 +0300
@@ -36,6 +36,7 @@
 
 	unsigned int last_indexed_uid_set:1;
 	unsigned int headers_open:1;
+	unsigned int body_open:1;
 	unsigned int cur_header_index:1;
 	unsigned int documents_added:1;
 	unsigned int expunges:1;
@@ -260,6 +261,10 @@
 fts_backend_solr_doc_close(struct solr_fts_backend_update_context *ctx)
 {
 	ctx->headers_open = FALSE;
+	if (ctx->body_open) {
+		ctx->body_open = FALSE;
+		str_append(ctx->cmd, "</field>");
+	}
 	if (str_len(ctx->hdr) > 0) {
 		str_append(ctx->cmd, "<field name=\"hdr\">");
 		str_append_str(ctx->cmd, ctx->hdr);
@@ -416,7 +421,10 @@
 		break;
 	case FTS_BACKEND_BUILD_KEY_BODY_PART:
 		ctx->headers_open = FALSE;
-		str_append(ctx->cmd, "<field name=\"body\">");
+		if (!ctx->body_open) {
+			ctx->body_open = TRUE;
+			str_append(ctx->cmd, "<field name=\"body\">");
+		}
 		break;
 	case FTS_BACKEND_BUILD_KEY_BODY_PART_BINARY:
 		i_unreached();
@@ -430,12 +438,15 @@
 	struct solr_fts_backend_update_context *ctx =
 		(struct solr_fts_backend_update_context *)_ctx;
 
-	if (!ctx->headers_open)
-		str_append(ctx->cmd, "</field>");
-	else {
+	if (ctx->headers_open) {
 		/* this is called individually for each header line.
 		   headers are finished only when key changes to body */
 		str_append_c(ctx->hdr, '\n');
+	} else {
+		i_assert(ctx->body_open);
+		/* messages can have multiple MIME bodies.
+		   add them all as one. */
+		str_append_c(ctx->cmd, '\n');
 	}
 
 	if (ctx->cur_header_index) {