From 2ffdbf50740a5e5124c72fec0252b6fd71add8cf Mon Sep 17 00:00:00 2001
From: John Kerl <kerl.john.r@gmail.com>
Date: Wed, 1 Feb 2017 21:22:13 -0500
Subject: [PATCH] neaten

---
 c/input/lrec_reader_mmap_json.c  | 18 ++++++++++--------
 c/input/lrec_reader_mmap_xtab.c  | 29 +++++++++++++++++++++++++----
 c/input/lrec_reader_stdio_xtab.c |  2 --
 3 files changed, 35 insertions(+), 14 deletions(-)

diff --git a/c/input/lrec_reader_mmap_json.c b/c/input/lrec_reader_mmap_json.c
index afd422567..582163049 100644
--- a/c/input/lrec_reader_mmap_json.c
+++ b/c/input/lrec_reader_mmap_json.c
@@ -128,15 +128,17 @@ static void lrec_reader_mmap_json_sof(void* pvstate, void* pvhandle) {
 	while (TRUE) {
 
 		// Find the first line-ending sequence (if any): LF or CRLF.
-		if (detected_line_term == NULL) {
-			for (char* p = phandle->sol; p < phandle->eof; p++) {
-				if (p[0] == '\n') {
-					if (p > phandle->sol && p[-1] == '\r') {
-						detected_line_term = "\r\n";
-					} else {
-						detected_line_term = "\n";
+		if (pstate->do_auto_line_term) {
+			if (detected_line_term == NULL) {
+				for (char* p = phandle->sol; p < phandle->eof; p++) {
+					if (p[0] == '\n') {
+						if (p > phandle->sol && p[-1] == '\r') {
+							detected_line_term = "\r\n";
+						} else {
+							detected_line_term = "\n";
+						}
+						break;
 					}
-					break;
 				}
 			}
 		}
diff --git a/c/input/lrec_reader_mmap_xtab.c b/c/input/lrec_reader_mmap_xtab.c
index 28c02f67a..9c1a448b7 100644
--- a/c/input/lrec_reader_mmap_xtab.c
+++ b/c/input/lrec_reader_mmap_xtab.c
@@ -126,8 +126,8 @@ static lrec_t* lrec_reader_mmap_xtab_process_multi_ifs_multi_ips(void* pvstate,
 lrec_t* lrec_parse_mmap_xtab_single_ifs_single_ips(file_reader_mmap_state_t* phandle, char ifs, char ips,
 	int allow_repeat_ips, int do_auto_line_term, context_t* pctx)
 {
-	// xxx comment all this
 	if (do_auto_line_term) {
+		// Skip over otherwise empty LF-only or CRLF-only lines.
 		while (phandle->sol < phandle->eof) {
 			if (*phandle->sol == '\n') {
 				context_set_autodetected_lf(pctx);
@@ -145,6 +145,7 @@ lrec_t* lrec_parse_mmap_xtab_single_ifs_single_ips(file_reader_mmap_state_t* pha
 			}
 		}
 	} else {
+		// Skip over otherwise empty IRS-only lines.
 		while (phandle->sol < phandle->eof && *phandle->sol == ifs)
 			phandle->sol++;
 	}
@@ -235,9 +236,29 @@ lrec_t* lrec_parse_mmap_xtab_single_ifs_single_ips(file_reader_mmap_state_t* pha
 lrec_t* lrec_parse_mmap_xtab_single_ifs_multi_ips(file_reader_mmap_state_t* phandle, char ifs, char* ips, int ipslen,
 	int allow_repeat_ips, int do_auto_line_term, context_t* pctx)
 {
-	// xxx autolineterm ....
-	while (phandle->sol < phandle->eof && *phandle->sol == ifs)
-		phandle->sol++;
+	if (do_auto_line_term) {
+		// Skip over otherwise empty LF-only or CRLF-only lines.
+		while (phandle->sol < phandle->eof) {
+			if (*phandle->sol == '\n') {
+				context_set_autodetected_lf(pctx);
+				phandle->sol += 1;
+			} else if (*phandle->sol == '\r') {
+				char* q = phandle->sol + 1;
+				if (q < phandle->eof && *q == '\n') {
+					context_set_autodetected_crlf(pctx);
+					phandle->sol += 2;
+				} else {
+					phandle->sol += 1;
+				}
+			} else {
+				break;
+			}
+		}
+	} else {
+		// Skip over otherwise empty IRS-only lines.
+		while (phandle->sol < phandle->eof && *phandle->sol == ifs)
+			phandle->sol++;
+	}
 
 	if (phandle->sol >= phandle->eof)
 		return NULL;
diff --git a/c/input/lrec_reader_stdio_xtab.c b/c/input/lrec_reader_stdio_xtab.c
index 755a9f939..528c940e9 100644
--- a/c/input/lrec_reader_stdio_xtab.c
+++ b/c/input/lrec_reader_stdio_xtab.c
@@ -79,7 +79,6 @@ static lrec_t* lrec_reader_stdio_xtab_process(void* pvstate, void* pvhandle, con
 
 	while (TRUE) {
 		int line_length = 0;
-		// xxx move all this into mlr_get_cline_auto_line_term ?
 		char* line = (pstate->ifslen == 1)
 			? mlr_get_cline_with_length(input_stream, pstate->ifs[0], &line_length)
 			: mlr_get_sline(input_stream, pstate->ifs, pstate->ifslen);
@@ -106,7 +105,6 @@ static lrec_t* lrec_reader_stdio_xtab_process(void* pvstate, void* pvhandle, con
 						pstate->allow_repeat_ips);
 			}
 
-		// xxx simplify
 		} else if (line[0] == '\r' && line[1] == '\0') {
 			free(line);
 			context_set_autodetected_crlf(pctx);