From 69df0b216fc51889aee3323a5505b7d49ffc5b4d Mon Sep 17 00:00:00 2001
From: Andy C <andy@oilshell.org>
Date: Tue, 14 Jan 2025 11:56:31 -0500
Subject: [PATCH] [refactor] Move files out of lazylex/

Preparing for data_lang/htm8.py.

Add htm8.asdl.  Not used yet.
---
 build/py.sh                                   |  1 +
 data_lang/htm8.asdl                           | 24 +++++++++++++++++++
 .../testdata/hello.htm8                       |  0
 lazylex/README.md => doc/lazylex.md           |  0
 doctools/oils_doc_test.py                     |  4 ++--
 lazylex/html_test.py                          |  2 +-
 6 files changed, 28 insertions(+), 3 deletions(-)
 create mode 100644 data_lang/htm8.asdl
 rename lazylex/testdata.html => data_lang/testdata/hello.htm8 (100%)
 rename lazylex/README.md => doc/lazylex.md (100%)

diff --git a/build/py.sh b/build/py.sh
index 625a376965..9a1e91170d 100755
--- a/build/py.sh
+++ b/build/py.sh
@@ -139,6 +139,7 @@ py-codegen() {
   # depends on syntax.asdl
   gen-asdl-py 'core/runtime.asdl'
   gen-asdl-py 'core/value.asdl'
+  gen-asdl-py 'data_lang/htm8.asdl'
   gen-asdl-py 'data_lang/nil8.asdl'
   gen-asdl-py 'display/pretty.asdl'
   gen-asdl-py 'mycpp/mycpp.asdl'
diff --git a/data_lang/htm8.asdl b/data_lang/htm8.asdl
new file mode 100644
index 0000000000..70bdbeae08
--- /dev/null
+++ b/data_lang/htm8.asdl
@@ -0,0 +1,24 @@
+
+module htm8
+{
+
+  h8_id =
+    Decl 
+
+    # CommentBegin, ProcessingBegin, CDataBegin are "pseudo-tokens", not visible
+  | Comment | CommentBegin 
+  | Processing | ProcessingBegin
+  | CData | CDataBegin
+
+  | StartTag | StartEndTag | EndTag 
+
+  | DecChar | HexChar | CharEntity 
+
+  | RawData | HtmlCData 
+
+  | BadAmpersand | BadGreaterThan | BadLessThan 
+
+  | Invalid
+  | EndOfStream
+  generate [no_namespace_suffix]  # cosmetic: call it h8_id, not h8_id_e
+}
diff --git a/lazylex/testdata.html b/data_lang/testdata/hello.htm8
similarity index 100%
rename from lazylex/testdata.html
rename to data_lang/testdata/hello.htm8
diff --git a/lazylex/README.md b/doc/lazylex.md
similarity index 100%
rename from lazylex/README.md
rename to doc/lazylex.md
diff --git a/doctools/oils_doc_test.py b/doctools/oils_doc_test.py
index c173cf17e9..14fc58a48f 100755
--- a/doctools/oils_doc_test.py
+++ b/doctools/oils_doc_test.py
@@ -7,7 +7,7 @@
 from lazylex import html
 from doctools import oils_doc  # module under test
 
-with open('lazylex/testdata.html') as f:
+with open('data_lang/testdata/hello.htm8') as f:
     TEST_HTML = f.read()
 
 
@@ -58,7 +58,7 @@ def testShPrompt(self):
 
     def testHighlightCode(self):
         # type: () -> None
-        # lazylex/testdata.html has the language-sh-prompt
+        # data_lang/testdata/hello.htm8 has the language-sh-prompt
 
         h = oils_doc.HighlightCode(TEST_HTML, None)
         self.assert_('<span class="sh-prompt">' in h, h)
diff --git a/lazylex/html_test.py b/lazylex/html_test.py
index 7cb9f65535..ebe0677b17 100755
--- a/lazylex/html_test.py
+++ b/lazylex/html_test.py
@@ -9,7 +9,7 @@
 
 log = html.log
 
-with open('lazylex/testdata.html') as f:
+with open('data_lang/testdata/hello.htm8') as f:
     TEST_HTML = f.read()