From c739b978a440bcc612566b627e73d1425340e7fb Mon Sep 17 00:00:00 2001
From: Nick Gasson <nick@nickg.me.uk>
Date: Wed, 20 Dec 2023 21:57:03 +0000
Subject: [PATCH] Avoid illegal characters in design unit file names. Fixes
 #816

---
 src/lib.c                      | 42 ++++++++++++++++++++++------
 test/regress/gold/issue816.txt |  1 +
 test/regress/issue816.vhd      | 51 ++++++++++++++++++++++++++++++++++
 test/regress/testlist.txt      |  1 +
 4 files changed, 86 insertions(+), 9 deletions(-)
 create mode 100644 test/regress/gold/issue816.txt
 create mode 100644 test/regress/issue816.vhd

diff --git a/src/lib.c b/src/lib.c
index d4c3d5ae..cc48e964 100644
--- a/src/lib.c
+++ b/src/lib.c
@@ -736,15 +736,37 @@ void lib_put_error(lib_t lib, tree_t unit)
    lib_put_aux(lib, obj, true, true, get_real_time(), NULL);
 }
 
-static lib_unit_t *lib_read_unit(lib_t lib, const char *fname)
+static void lib_encode_file_name(ident_t id, text_buf_t *tb)
 {
-   fbuf_t *f = lib_fbuf_open(lib, fname, FBUF_IN, FBUF_CS_ADLER32);
+   // Encode extended identifiers in file names using the hexadecimal
+   // value of the ISO-8859-1 code points to avoid illegal characters
+   // and case sensitivity issues on Windows and macOS
+
+   bool esc = false;
+   for (const char *p = istr(id); *p; p++) {
+      if (*p == '\\') {
+         esc = !esc;
+         tb_append(tb, '+');
+      }
+      else if (esc)
+         tb_printf(tb, "%02x", *p);
+      else
+         tb_append(tb, *p);
+   }
+}
+
+static lib_unit_t *lib_read_unit(lib_t lib, ident_t id)
+{
+   LOCAL_TEXT_BUF tb = tb_new();
+   lib_encode_file_name(id, tb);
+
+   fbuf_t *f = lib_fbuf_open(lib, tb_get(tb), FBUF_IN, FBUF_CS_ADLER32);
    if (f == NULL)
       return NULL;
 
    file_info_t info;
    if (!get_handle_info(fbuf_file_handle(f), &info))
-      fatal_errno("%s", fname);
+      fatal_errno("%s", tb_get(tb));
 
    ident_rd_ctx_t ident_ctx = ident_read_begin(f);
    loc_rd_ctx_t *loc_ctx = loc_read_begin(f);
@@ -762,7 +784,7 @@ static lib_unit_t *lib_read_unit(lib_t lib, const char *fname)
          // Ignore it (remove after 1.12 release)
          break;
       default:
-         fatal_trace("unhandled tag %c in %s", tag, fname);
+         fatal_trace("unhandled tag %c in %s", tag, tb_get(tb));
       }
    }
 
@@ -773,7 +795,7 @@ static lib_unit_t *lib_read_unit(lib_t lib, const char *fname)
    fbuf_close(f, &checksum);
 
    if (obj == NULL)
-      fatal_trace("%s did not HDL design unit", fname);
+      fatal_trace("%s did not contain a HDL design unit", tb_get(tb));
 
    arena_set_checksum(object_arena(obj), checksum);
 
@@ -802,7 +824,7 @@ static lib_unit_t *lib_get_aux(lib_t lib, ident_t ident)
    file_read_lock(lib->lock_fd);
 
    // Otherwise search in the filesystem if not in the cache
-   lu = lib_read_unit(lib, istr(ident));
+   lu = lib_read_unit(lib, ident);
 
    file_unlock(lib->lock_fd);
 
@@ -940,10 +962,12 @@ ident_t lib_name(lib_t lib)
 
 static void lib_save_unit(lib_t lib, lib_unit_t *unit)
 {
-   fbuf_t *f = lib_fbuf_open(lib, istr(unit->name), FBUF_OUT, FBUF_CS_ADLER32);
+   LOCAL_TEXT_BUF tb = tb_new();
+   lib_encode_file_name(unit->name, tb);
+
+   fbuf_t *f = lib_fbuf_open(lib, tb_get(tb), FBUF_OUT, FBUF_CS_ADLER32);
    if (f == NULL)
-      fatal("failed to create %s in library %s", istr(unit->name),
-            istr(lib->name));
+      fatal("failed to create %s in library %s", tb_get(tb), istr(lib->name));
 
    write_u8('T', f);
 
diff --git a/test/regress/gold/issue816.txt b/test/regress/gold/issue816.txt
new file mode 100644
index 00000000..3398f2a9
--- /dev/null
+++ b/test/regress/gold/issue816.txt
@@ -0,0 +1 @@
+2ns+0: Report Failure: passed
diff --git a/test/regress/issue816.vhd b/test/regress/issue816.vhd
new file mode 100644
index 00000000..c3a55b9a
--- /dev/null
+++ b/test/regress/issue816.vhd
@@ -0,0 +1,51 @@
+entity \foo\ is
+    port ( x : out integer );
+end entity;
+
+architecture \test\ of \foo\ is
+begin
+    x <= 1;
+end architecture;
+
+-------------------------------------------------------------------------------
+
+entity \Foo\ is
+    port ( x : out integer );
+end entity;
+
+architecture \test\ of \Foo\ is
+begin
+    x <= 2;
+end architecture;
+
+-------------------------------------------------------------------------------
+
+entity issue816 is
+end entity;
+
+architecture test of issue816 is
+    signal a, b : integer;
+begin
+
+    u1: entity work.\foo\ port map ( a );
+
+    u2: entity work.\Foo\ port map ( b );
+
+    check: process is
+    begin
+        wait for 1 ns;
+
+        assert a = 1 severity failure;
+        assert b = 2 severity failure;
+
+        wait for 1 ns;
+
+        -- Make this test fail so that run_regr does analysis/elaboration and
+        -- execution in two separate processes, causing the library units to be
+        -- re-read from disk
+        report "passed" severity failure;
+
+        wait;
+    end process;
+
+end architecture;
diff --git a/test/regress/testlist.txt b/test/regress/testlist.txt
index ad87b260..2466eba4 100644
--- a/test/regress/testlist.txt
+++ b/test/regress/testlist.txt
@@ -913,3 +913,4 @@ cmdline9        shell,slow
 genpack16       normal,2008
 issue815        normal,2019
 issue817        fail,gold,2008
+issue816        fail,gold
-- 
2.39.2