From fba3371d4c82c95433ac1f09c1b498eaa8dd67d0 Mon Sep 17 00:00:00 2001
From: Thorsten Töpper <atsutane@freethoughts.de>
Date: Sat, 30 Aug 2025 18:12:28 +0200
Subject: split_for_sort: handle filename - as stdin

---
 src/split_for_sort.c | 132 ++++++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 125 insertions(+), 7 deletions(-)

diff --git a/src/split_for_sort.c b/src/split_for_sort.c
index 5678f6d..68b7819 100644
--- a/src/split_for_sort.c
+++ b/src/split_for_sort.c
@@ -45,6 +45,7 @@ struct list_head {
 int push_into_list_unique(struct list_head *list, char *name);
 int split_into_files(struct list_head *list, char *inputfile, char *prefix);
 size_t list_check_length(struct list_head *list);
+struct list_head *stdin_handling(struct list_head *list, size_t splitlength, char *prefix);
 struct list_head *extend_list(struct list_head *list, size_t splitlength, char *fname);
 struct list_node *get_node(struct list_head *list, char *name);
 void destroy_list(struct list_head *list);
@@ -373,11 +374,116 @@ int split_into_files(struct list_head *list, char *inputfile, char *prefix) {
     return 0;
 }
 
+/* This covers extend_list and split_into_files for the stdin stream.
+ * The other two functions can't be used as extend_list would read
+ * the data from stdin and split_into_files would have nothing to work with.
+ */
+struct list_head *stdin_handling(struct list_head *list, size_t splitlength, char *prefix) {
+    FILE *fdin = stdin;
+    char *filename = NULL, *line = NULL, *line_head = NULL, *previous = NULL;
+    size_t i=0, len = 0;
+    struct list_node *node = NULL;
+
+    if ((line = calloc(BUFFERSIZE, sizeof(char))) == NULL) {
+        LOGERR("ERROR: Failed to allocate memory for read buffer\n");
+        return NULL;
+    }
+    if ((line_head = calloc(BUFFERSIZE, sizeof(char))) == NULL) {
+        LOGERR("ERROR: Failed to allocate memory for read buffer\n");
+        free(line);
+        return NULL;
+    }
+
+    if ((previous = calloc(BUFFERSIZE, sizeof(char))) == NULL) {
+        LOGERR("ERROR: Failed to allocate memory for read buffer\n");
+        free(line);
+        free(line_head);
+        return NULL;
+    }
+    if ((filename = calloc(BUFFERSIZE, sizeof(char))) == NULL) {
+        LOGERR("ERROR: Failed to allocate memory for input buffer.\n");
+        free(line);
+        free(line_head);
+        free(previous);
+        return NULL;
+    }
+
+    if ((list == NULL) && \
+        ((list = calloc(1, sizeof(struct list_head))) == NULL)) {
+            LOGERR("ERROR: Failed to create new list");
+            free(line);
+            free(line_head);
+            free(previous);
+            free(filename);
+            return NULL;
+    }
+    list->splitlength = splitlength;
+
+    while (fgets(line, BUFFERSIZE, fdin) != NULL) {
+        len = strlen(line);
+        len = (splitlength < len) ? splitlength : len;
+        /* most simple way to stick with FS compatible characters */
+        for (i=0; i<len; i++) {
+            if (! isalnum(line[i]) ) {
+                line_head[i] = '_';
+            } else {
+                line_head[i] = line[i];
+            }
+        }
+        line_head[len] = '\0';
+
+        if (strncmp(line_head, previous, splitlength) != 0) {
+            if (push_into_list_unique(list, line_head) != 0) {
+                LOGERR("WARNING: Failed to insert '%s' into list. (file %s)\n", line, filename);
+                continue;
+            }
+
+            node = get_node(list, line_head);
+            if (node == NULL) {
+                LOGERR("WARNING: no node found for %s - skip line\n", line_head);
+                continue;
+            }
+
+            /* keep the file descriptors open across the input files to be split */
+            if (node->fd == NULL) {
+                snprintf(filename, BUFFERSIZE, "%s%s", ((prefix==NULL)?"":prefix), node->name);
+                if ((node->fd = fopen(filename, (option_append_mode)?"a":"w")) == NULL) {
+                    LOGERR("ERROR: Failed to open file '%s': %s\n", filename, strerror(errno));
+                    free(line);
+                    free(line_head);
+                    free(previous);
+                    free(filename);
+                    destroy_list(list);
+                    return NULL;
+                }
+            }
+            strncpy(previous, line_head, splitlength);
+        }
+
+        if (fputs(line, node->fd) == EOF) {
+            LOGERR("ERROR: Failed to write into output file for '%s': %s\n",
+                    line_head, strerror(errno));
+            free(line);
+            free(line_head);
+            free(previous);
+            free(filename);
+            destroy_list(list);
+            return NULL;
+        }
+    }
+
+    free(line);
+    free(line_head);
+    free(previous);
+    free(filename);
+
+    return list;
+}
 
 int main(int argc, char **argv) {
     int data_index = 2, length_index = 1, opt = 0, output_index = 0;
     size_t splitlength = 0;
-    struct list_head *list = NULL;
+    struct list_head *list = NULL, *lsttmp = NULL;
 
     if (argc < 4) {
         fprintf(stderr, "Usage: %s prefix length files...\n", argv[0]);
@@ -411,12 +517,24 @@ int main(int argc, char **argv) {
     }
 
     for (; data_index<argc; data_index++) {
-//        fprintf(stderr, "=> Scan '%s' for bucket filenames\n", argv[i]);
-        list = extend_list(list, splitlength, argv[data_index]);
-//        fprintf(stderr,"=> Split '%s' into bucket files\n", argv[i]);
-        if (split_into_files(list, argv[data_index], argv[output_index]) < 0) {
-            destroy_list(list);
-            return EXIT_FAILURE;
+        if (argv[data_index][0] == '-' && argv[data_index][1] == '\0') {
+            lsttmp = stdin_handling(list, splitlength, argv[output_index]);
+            if (lsttmp == NULL) {
+                destroy_list(list);
+                return EXIT_FAILURE;
+            }
+            list = lsttmp;
+        } else {
+            lsttmp = extend_list(list, splitlength, argv[data_index]);
+            if (lsttmp == NULL) {
+                destroy_list(list);
+                return EXIT_FAILURE;
+            }
+            list = lsttmp;
+            if (split_into_files(list, argv[data_index], argv[output_index]) < 0) {
+                destroy_list(list);
+                return EXIT_FAILURE;
+            }
         }
     }
 
-- 
cgit v1.2.3-70-g09d2