Add (/path):by_line()

6 files changed, 192 insertions, 2 deletions

diff --git a/builtins/nextline.c b/builtins/nextline.c
new file mode 100644
index 00000000..3fa37412
--- /dev/null
+++ b/builtins/nextline.c
@@ -0,0 +1,108 @@
+// An enum used for iterating over lines in a file
+// Most of the code here was generated by compiling:
+//    enum NextLine(Done, Next(line:Text))
+
+#include <stdbool.h>
+#include <stdint.h>
+
+#include "siphash.h"
+#include "datatypes.h"
+#include "nextline.h"
+#include "text.h"
+#include "util.h"
+
+static Text_t NextLine$Next$as_text(NextLine$Next_t *obj, bool use_color)
+{
+    if (!obj)
+        return Text("Next");
+    return Text$concat(use_color ? Text("\x1b[0;1mNext\x1b[m(") : Text("Next("), Text("line="),
+                       Text$as_text((Text_t[1]){obj->$line}, use_color, &Text$info), Text(")"));
+}
+
+public inline NextLine_t NextLine$tagged$Next(Text_t $line)
+{
+    return (NextLine_t) {
+        .tag = NextLine$tag$Next,.$Next = { $line }
+    };
+}
+
+static Text_t NextLine$as_text(NextLine_t *obj, bool use_color)
+{
+    if (!obj)
+        return Text("NextLine");
+    switch (obj->tag) {
+    case NextLine$tag$Done:
+        return use_color ? Text("\x1b[36;1mNextLine.Done\x1b[m") : Text("NextLine.Done");
+    case NextLine$tag$Next:
+        return Text$concat(use_color ? Text("\x1b[36;1mNextLine.Next\x1b[m(") :
+                           Text("NextLine.Next("), Text("line="),
+                           Text$as_text((Text_t[1]){obj->$Next.$line}, use_color, &Text$info), Text(")"));
+    default:
+        return (Text_t) {
+        .length = 0};
+    }
+}
+
+static bool NextLine$equal(const NextLine_t *x, const NextLine_t *y,
+                               const TypeInfo *info)
+{
+    (void) info;
+    if (x->tag != y->tag)
+        return false;
+    switch (x->tag) {
+    case NextLine$tag$Done:
+        return false;
+    case NextLine$tag$Next:
+        return generic_equal(&x->$Next, &y->$Next, (&NextLine$Next));
+    default:
+        return 0;
+    }
+}
+
+static int NextLine$compare(const NextLine_t *x, const NextLine_t *y,
+                                const TypeInfo *info)
+{
+    (void) info;
+    int diff = (int)x->tag - (int)y->tag;
+    if (diff)
+        return diff;
+    switch (x->tag) {
+    case NextLine$tag$Done:
+        return 0;
+    case NextLine$tag$Next:
+        return generic_compare(&x->$Next, &y->$Next, (&NextLine$Next));
+    default:
+        return 0;
+    }
+}
+
+static uint64_t NextLine$hash(const NextLine_t *obj, const TypeInfo *info)
+{
+    (void) info;
+    uint64_t hashes[2] = { (uint64_t) obj->tag, 0 };
+    switch (obj->tag) {
+    case NextLine$tag$Done:
+        break;
+    case NextLine$tag$Next:
+        hashes[1] = generic_hash(&obj->$Next, (&NextLine$Next));
+        break;
+    default: break;
+    }
+    return siphash24((void *) &hashes, sizeof(hashes));
+}
+
+public const TypeInfo NextLine$Done = { 0, 0, {.tag = EmptyStruct,.EmptyStruct.name =
+                                                   "NextLine$Done" } };
+public const TypeInfo NextLine$Next = { 24, 8, {.tag = CustomInfo,.CustomInfo =
+                                                    {.as_text =
+                                                     (void *) NextLine$Next$as_text,.hash =
+                                                     (void *) Text$hash,.compare =
+                                                     (void *) Text$compare,.equal =
+                                                     (void *) Text$equal,} } };
+public const TypeInfo NextLine = { 32, 8, {.tag = CustomInfo,.CustomInfo =
+                                               {.as_text = (void *) NextLine$as_text,.equal =
+                                                (void *) NextLine$equal,.hash =
+                                                (void *) NextLine$hash,.compare =
+                                                (void *) NextLine$compare} } };
+
+// vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1,\:0
diff --git a/builtins/nextline.h b/builtins/nextline.h
new file mode 100644
index 00000000..ce4c9fab
--- /dev/null
+++ b/builtins/nextline.h
@@ -0,0 +1,31 @@
+#pragma once
+
+#include "datatypes.h"
+#include "types.h"
+
+// An enum used for iterating over lines in a file
+// Most of the code here was generated by compiling:
+//    enum NextLine(Done, Next(line:Text))
+
+typedef struct NextLine_s NextLine_t;
+typedef struct NextLine$Done_s NextLine$Done_t;
+#pragma GCC diagnostic ignored "-Wpedantic"
+struct NextLine$Done_s {
+};
+typedef struct NextLine$Next_s NextLine$Next_t;
+struct NextLine$Next_s {
+    Text_t $line;
+};
+struct NextLine_s {
+    enum { NextLine$tag$Done = 0, NextLine$tag$Next = 1 } tag;
+    union {
+        NextLine$Done_t $Done;
+        NextLine$Next_t $Next;
+    };
+};
+extern const TypeInfo NextLine;
+extern const TypeInfo NextLine$Done;
+extern const TypeInfo NextLine$Next;
+NextLine_t NextLine$tagged$Next(Text_t $line);
+
+// vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1,\:0
diff --git a/builtins/path.c b/builtins/path.c
index 7c2198cf..096cbdd0 100644
--- a/builtins/path.c
+++ b/builtins/path.c
@@ -1,10 +1,11 @@
 // A lang for filesystem paths
 #include <dirent.h>
-#include <fcntl.h>
 #include <errno.h>
-#include <string.h>
+#include <fcntl.h>
+#include <gc.h>
 #include <stdbool.h>
 #include <stdint.h>
+#include <string.h>
 #include <sys/mman.h>
 #include <sys/stat.h>
 #include <sys/types.h>
@@ -15,6 +16,7 @@
 #include "files.h"
 #include "functions.h"
 #include "integers.h"
+#include "nextline.h"
 #include "path.h"
 #include "text.h"
 #include "types.h"
@@ -423,6 +425,51 @@ public Text_t Path$extension(Path_t path, bool full)
         return Text("");
 }
 
+static void _line_reader_cleanup(FILE **f)
+{
+    if (f && *f) {
+        fclose(*f);
+        *f = NULL;
+    }
+}
+
+static NextLine_t _next_line(FILE **f)
+{
+    if (!f || !*f) return (NextLine_t){NextLine$tag$Done};
+
+    char *line = NULL;
+    size_t size = 0;
+    ssize_t len = getline(&line, &size, *f);
+    if (len <= 0) {
+        _line_reader_cleanup(f);
+        return (NextLine_t){NextLine$tag$Done};
+    }
+
+    while (len > 0 && (line[len-1] == '\r' || line[len-1] == '\n'))
+        --len;
+
+    if (u8_check((uint8_t*)line, (size_t)len) != NULL)
+        fail("Invalid UTF8!");
+
+    Text_t line_text = Text$format("%.*s", len, line);
+    free(line);
+    return NextLine$tagged$Next(line_text);
+}
+
+public closure_t Path$by_line(Path_t path)
+{
+    path = Path$_expand_home(path);
+
+    FILE *f = fopen(Text$as_c_string(path), "r");
+    if (f == NULL)
+        fail("Could not read file: %k (%s)", &path, strerror(errno));
+
+    FILE **wrapper = GC_MALLOC(sizeof(FILE*));
+    *wrapper = f;
+    GC_register_finalizer(wrapper, (void*)_line_reader_cleanup, NULL, NULL, NULL);
+    return (closure_t){.fn=(void*)_next_line, .userdata=wrapper};
+}
+
 public const TypeInfo Path$info = {
     .size=sizeof(Path_t),
     .align=__alignof__(Path_t),
@@ -430,4 +477,5 @@ public const TypeInfo Path$info = {
     .TextInfo={.lang="Path"},
 };
 
+
 // vim: ts=4 sw=0 et cino=L2,l1,(0,W4,m1,\:0
diff --git a/builtins/path.h b/builtins/path.h
index 6772f72d..dd42af85 100644
--- a/builtins/path.h
+++ b/builtins/path.h
@@ -39,6 +39,7 @@ Text_t Path$write_unique(Path_t path, Text_t text);
 Path_t Path$parent(Path_t path);
 Text_t Path$base_name(Path_t path);
 Text_t Path$extension(Path_t path, bool full);
+closure_t Path$by_line(Path_t path);
 
 #define Path$hash Text$hash
 #define Path$compare Text$compare
diff --git a/builtins/siphash.h b/builtins/siphash.h
index a466200c..0fc7614c 100644
--- a/builtins/siphash.h
+++ b/builtins/siphash.h
@@ -3,6 +3,7 @@
 // An implementation of the SipHash algorithm.
 
 #include <stdint.h>
+#include <stddef.h>
 
 uint64_t siphash24(const uint8_t *src, size_t src_sz);
 
diff --git a/builtins/tomo.h b/builtins/tomo.h
index c7814376..29f71f71 100644
--- a/builtins/tomo.h
+++ b/builtins/tomo.h
@@ -17,6 +17,7 @@
 #include "datatypes.h"
 #include "functions.h"
 #include "integers.h"
+#include "nextline.h"
 #include "macros.h"
 #include "memory.h"
 #include "nums.h"