From 705ad09de77a299f83eaaf91572f71fe8ae77765 Mon Sep 17 00:00:00 2001
From: Jeremy Howard <github@jhoward.fastmail.fm>
Date: Sun, 27 Oct 2024 18:51:06 +1000
Subject: [PATCH] fixes #16

---
 00_xml.ipynb   | 90 +++++++++++++++++---------------------------------
 toolslm/xml.py |  6 ++--
 2 files changed, 33 insertions(+), 63 deletions(-)

diff --git a/00_xml.ipynb b/00_xml.ipynb
index c049b0e..065b433 100644
--- a/00_xml.ipynb
+++ b/00_xml.ipynb
@@ -225,7 +225,7 @@
     {
      "data": {
       "text/plain": [
-       "doctype(source='\\nb8898fab\\n', content='\\nThis is a sample\\n')"
+       "doctype(source='\\n47e19350\\n', content='\\nThis is a \"sample\"\\n')"
       ]
      },
      "execution_count": null,
@@ -234,7 +234,7 @@
     }
    ],
    "source": [
-    "doc = 'This is a sample'\n",
+    "doc = 'This is a \"sample\"'\n",
     "mk_doctype(doc)"
    ]
   },
@@ -253,8 +253,8 @@
     "          ) -> tuple:\n",
     "    \"Create an `ft` format tuple for a single doc in Anthropic's recommended format\"\n",
     "    dt = mk_doctype(content, source)\n",
-    "    content = Document_content(dt.content)\n",
-    "    source = Source(dt.source)\n",
+    "    content = Document_content(NotStr(dt.content))\n",
+    "    source = Source(NotStr(dt.source))\n",
     "    return Document(source, content, index=index, **kwargs)"
    ]
   },
@@ -276,19 +276,15 @@
      "data": {
       "text/markdown": [
        "```html\n",
-       "<document index=\"1\" title=\"test\">\n",
-       "  <source>\n",
-       "b8898fab\n",
-       "\n",
-       "  <document-content>\n",
-       "This is a sample\n",
-       "</document-content>\n",
-       "</document>\n",
-       "\n",
+       "<document index=\"1\" title=\"test\"><source>\n",
+       "47e19350\n",
+       "<document-content>\n",
+       "This is a \"sample\"\n",
+       "</document-content></document>\n",
        "```"
       ],
       "text/plain": [
-       "document((source(('\\nb8898fab\\n',),{}), document-content(('\\nThis is a sample\\n',),{})),{'index': 1, 'title': 'test'})"
+       "document((source(('\\n47e19350\\n',),{}), document-content(('\\nThis is a \"sample\"\\n',),{})),{'index': 1, 'title': 'test'})"
       ]
      },
      "execution_count": null,
@@ -341,25 +337,15 @@
      "text": [
       "Here are some documents for you to reference for your task:\n",
       "\n",
-      "<documents>\n",
-      "  <document index=\"1\">\n",
-      "    <source>\n",
-      "b8898fab\n",
-      "\n",
-      "    <document-content>\n",
-      "This is a sample\n",
-      "</document-content>\n",
-      "  </document>\n",
-      "  <document index=\"2\">\n",
-      "    <source>\n",
+      "<documents><document index=\"1\"><source>\n",
+      "47e19350\n",
+      "<document-content>\n",
+      "This is a \"sample\"\n",
+      "</document-content></document><document index=\"2\"><source>\n",
       "doc.txt\n",
-      "\n",
-      "    <document-content>\n",
+      "<document-content>\n",
       "And another one\n",
-      "</document-content>\n",
-      "  </document>\n",
-      "</documents>\n",
-      "\n"
+      "</document-content></document></documents>\n"
      ]
     }
    ],
@@ -430,29 +416,19 @@
        "```xml\n",
        "Here are some documents for you to reference for your task:\n",
        "\n",
-       "<documents>\n",
-       "  <document index=\"1\">\n",
-       "    <source>\n",
+       "<documents><document index=\"1\"><source>\n",
        "samples/sample_core.py\n",
-       "\n",
-       "    <document-content>\n",
+       "<document-content>\n",
        "import inspect\n",
        "empty = inspect.Parameter.empty\n",
-       "models = &#x27;claude-3-opus-20240229&#x27;,&#x27;claude-3-sonnet-20240229&#x27;,&#x27;claude-3-haiku-20240307&#x27;\n",
-       "</document-content>\n",
-       "  </document>\n",
-       "  <document index=\"2\">\n",
-       "    <source>\n",
+       "models = 'claude-3-opus-20240229','claude-3-sonnet-20240229','claude-3-haiku-20240307'\n",
+       "</document-content></document><document index=\"2\"><source>\n",
        "samples/sample_styles.css\n",
-       "\n",
-       "    <document-content>\n",
+       "<document-content>\n",
        ".cell { margin-bottom: 1rem; }\n",
-       ".cell &gt; .sourceCode { margin-bottom: 0; }\n",
-       ".cell-output &gt; pre { margin-bottom: 0; }\n",
-       "</document-content>\n",
-       "  </document>\n",
-       "</documents>\n",
-       "\n",
+       ".cell > .sourceCode { margin-bottom: 0; }\n",
+       ".cell-output > pre { margin-bottom: 0; }\n",
+       "</document-content></document></documents>\n",
        "```"
       ],
       "text/plain": [
@@ -505,19 +481,13 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "<documents>\n",
-      "  <document index=\"1\">\n",
-      "    <source>\n",
+      "<documents><document index=\"1\"><source>\n",
       "samples/sample_core.py\n",
-      "\n",
-      "    <document-content>\n",
+      "<document-content>\n",
       "import inspect\n",
       "empty = inspect.Parameter.empty\n",
-      "models = &#x27;claude-3-opus-20240229&#x27;,&#x27;claude-3-sonnet-20240229&#x27;,&#x27;claude-3-haiku-20240307&#x27;\n",
-      "</document-content>\n",
-      "  </document>\n",
-      "</documents>\n",
-      "\n"
+      "models = 'claude-3-opus-20240229','claude-3-sonnet-20240229','claude-3-haiku-20240307'\n",
+      "</document-content></document></documents>\n"
      ]
     }
    ],
@@ -540,7 +510,7 @@
     "    folder:str, # Folder name containing files to add to context\n",
     "    **kwargs # Passed to `folder2ctx`\n",
     ")->str: # XML for Claude context\n",
-    "    return folder2ctx(folder, **kwargs)"
+    "    print(folder2ctx(folder, **kwargs))"
    ]
   },
   {
diff --git a/toolslm/xml.py b/toolslm/xml.py
index fff881f..4e6e8b7 100644
--- a/toolslm/xml.py
+++ b/toolslm/xml.py
@@ -58,8 +58,8 @@ def mk_doc(index:int,  # The document index
           ) -> tuple:
     "Create an `ft` format tuple for a single doc in Anthropic's recommended format"
     dt = mk_doctype(content, source)
-    content = Document_content(dt.content)
-    source = Source(dt.source)
+    content = Document_content(NotStr(dt.content))
+    source = Source(NotStr(dt.source))
     return Document(source, content, index=index, **kwargs)
 
 # %% ../00_xml.ipynb 19
@@ -101,4 +101,4 @@ def folder2ctx_cli(
     folder:str, # Folder name containing files to add to context
     **kwargs # Passed to `folder2ctx`
 )->str: # XML for Claude context
-    return folder2ctx(folder, **kwargs)
+    print(folder2ctx(folder, **kwargs))