From 9de405c3ea2756840f40c9296e9c992708a81675 Mon Sep 17 00:00:00 2001
From: liquidaty <info@liquidaty.com>
Date: Tue, 13 Aug 2024 13:25:25 -0700
Subject: [PATCH] fix cum_scanned_length tracking for files larger than initial
 read buff add zsv_peek()

---
 app/Makefile             |  2 +-
 app/builtin/help.c       |  2 +-
 app/select.c             |  3 ++-
 include/zsv/api.h        |  3 +++
 include/zsv/utils/dirs.h |  2 +-
 src/zsv.c                | 12 ++++++++++--
 6 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/app/Makefile b/app/Makefile
index bae75373..1b13ee53 100644
--- a/app/Makefile
+++ b/app/Makefile
@@ -260,7 +260,7 @@ MORE_SOURCE+= ${YAJL_INCLUDE} ${YAJL_HELPER_INCLUDE} -I${JQ_INCLUDE_DIR}
 MORE_LIBS+=${JQ_LIB} ${LDFLAGS_JQ}
 
 help:
-	@echo "To build: ${MAKE} [DEBUG=1] [clean] [clean-all] [BINDIR=${BINDIR}] [JQ_PREFIX=/usr/local] <install|all|test>"
+	@echo "To build: ${MAKE} [DEBUG=1] [clean] [clean-all] [BINDIR=${BINDIR}] [JQ_PREFIX=/usr/local] <install|all|install-util-lib|test>"
 	@echo
 	@echo "If JQ_PREFIX is not defined, libjq will be built in the build dir"
 	@echo
diff --git a/app/builtin/help.c b/app/builtin/help.c
index 277effd9..3b078a35 100644
--- a/app/builtin/help.c
+++ b/app/builtin/help.c
@@ -35,7 +35,7 @@ static int main_help(int argc, const char *argv[]) {
     "",
     "Options common to all commands except `prop`, `rm` and `jq`:",
 #ifdef ZSV_EXTRAS
-    "  -L,--limit-rows <n>: limit processing to the given number of rows (including any header row(s))",
+    "  -L,--limit-rows <n>      : limit processing to the given number of rows (including any header row(s))",
 #endif
     "  -c,--max-column-count <n>: set the maximum number of columns parsed per row. defaults to 1024",
     "  -r,--max-row-size <n>    : set the minimum supported maximum row size. defaults to 64k",
diff --git a/app/select.c b/app/select.c
index ea8618d8..340bf395 100644
--- a/app/select.c
+++ b/app/select.c
@@ -962,7 +962,8 @@ int ZSV_MAIN_FUNC(ZSV_COMMAND)(int argc, const char *argv[], struct zsv_opts *op
         while(status == zsv_status_ok
               && !zsv_signal_interrupted && !data.cancelled)
           status = zsv_parse_more(data.parser);
-        zsv_finish(data.parser);
+        if(status == zsv_status_no_more_input)
+          status = zsv_finish(data.parser);
         zsv_delete(data.parser);
       }
     }
diff --git a/include/zsv/api.h b/include/zsv/api.h
index 9bc05640..58ce3b1e 100644
--- a/include/zsv/api.h
+++ b/include/zsv/api.h
@@ -41,6 +41,9 @@
  * - zsv_delete(): dispose the parser
  ******************************************************************************/
 
+ZSV_EXPORT
+int zsv_peek(zsv_parser);
+
 /**
  * Create a zsv parser. Typically, passed options will at least include a
  * a `row_handler()` callback. Many, but not all, options can be subsequently
diff --git a/include/zsv/utils/dirs.h b/include/zsv/utils/dirs.h
index ed120300..d78b0886 100644
--- a/include/zsv/utils/dirs.h
+++ b/include/zsv/utils/dirs.h
@@ -73,7 +73,7 @@ struct zsv_foreach_dirent_handle {
 
   unsigned char verbose:1;
   unsigned char is_dir:1;       /* non-zero if this entry is a directory */
-  unsigned char no_recurse:1;        /* set to 1 when handling a dir to prevent recursing into it */
+  unsigned char no_recurse:1;   /* set to 1 when handling a dir to prevent recursing into it */
   unsigned char _:5;
 };
 
diff --git a/src/zsv.c b/src/zsv.c
index 10809d58..4f21fe1d 100644
--- a/src/zsv.c
+++ b/src/zsv.c
@@ -58,10 +58,11 @@ inline static size_t scanner_pre_parse(struct zsv_scanner *scanner) {
     scanner->old_bytes_read = 0;
   }
 
-  scanner->cum_scanned_length += scanner->scanned_length;
+  scanner->cum_scanned_length += scanner->scanned_length - scanner->partial_row_length;
 
   size_t capacity = scanner->buff.size - scanner->partial_row_length;
-  if(VERY_UNLIKELY(capacity == 0)) { // our row size was too small to fit a single row of data
+  if(VERY_UNLIKELY(capacity == 0)) {
+    // our row size was too small to fit a single row of data
     fprintf(stderr, "Warning: row %zu truncated\n", scanner->data_row_count);
     if(scanner->mode == ZSV_MODE_FIXED) {
       if(VERY_UNLIKELY(row_fx(scanner, scanner->buff.buff, 0, scanner->buff.size)))
@@ -310,6 +311,13 @@ ZSV_EXPORT enum zsv_status zsv_set_fixed_offsets(zsv_parser parser, size_t count
   return zsv_status_ok;
 }
 
+ZSV_EXPORT
+int zsv_peek(zsv_parser z) {
+  if(z->scanned_length + 1 < z->buff.size)
+    return z->buff.buff[z->scanned_length+1];
+  return -1;
+}
+
 /**
  * Create a zsv parser
  * @param opts