ziya 0.1.42__py3-none-any.whl → 0.1.44__py3-none-any.whl

app/agents/agent.py

@@ -55,16 +55,8 @@ def get_combined_docs_from_files(files) -> str:
         try:
             full_file_path = os.path.join(user_codebase_dir, file_path)
             docs = TextLoader(full_file_path).load()
-            for doc_index, doc in enumerate(docs):
-                # Add line numbers to the content
-                lines = doc.page_content.split('\n')
-                numbered_lines = [f"{i + 1:4d} | {line}" for i, line in enumerate(lines)]
-                numbered_content = '\n'.join(numbered_lines)
-                
-                # Use the numbered content instead of the original
-                doc.page_content = numbered_content
+            for doc in docs:
                 combined_contents += f"File: {file_path}\n{doc.page_content}\n\n"
-
         except Exception as e:
             print(f"Skipping file {full_file_path} due to error: {e}")
 

app/agents/prompts.py

@@ -6,10 +6,6 @@ template = """
 
 You are an excellent coder. Help the user with their coding tasks. You are given the codebase of the user in your context.
 
-IMPORTANT: The file contents provided to you have line numbers added at the beginning of each line in the format "   
-1 | <line content>". These line numbers start at 1, matching Git's conventions. Use these line numbers when generating 
-diff hunks to ensure accurate line numbering.
-
 IMPORTANT: When recommending code changes, format your response as a standard Git diff format unless the user specifies otherwise. 
 Follow these strict guidelines for diff formatting:
 
@@ -36,25 +32,23 @@ Follow these strict guidelines for diff formatting:
 5. End each diff block with ``` on a new line
 
 CRITICAL: After generating each hunk diff, carefully review and verify the following:
-1. Ensure all Hunk Headers (e.g., @@ -4,7 +4,2 @@) are accurate. The numbers should correctly reflect the lines being changed, added, or removed.
-2. Verify that the line numbers in the Hunk Headers match the actual content changes in the diff.
-3. Check that the diff can be applied cleanly using `git apply`. This means:
+1. Check that the diff can be applied cleanly using `git apply`. This means:
    - The context lines (unchanged lines) in the diff should match the original file content.
    - The line numbers and content should be consistent throughout the diff.
    - There should be no conflicts or inconsistencies in the changes.
-4. If you find any errors or inconsistencies, correct them before finalizing the diff.
-5. For each hunk in the diff, please make sure it starts or ends with a line containing content instead of empty line, if possible.
-6. When creating a new file, ensure the line `new file mode 100644` is included to specify file permissions.
+2. If you find any errors or inconsistencies, correct them before finalizing the diff.
+3. For each hunk in the diff, please make sure it starts or ends with a line containing content instead of empty line, if possible.
+4. When creating a new file, ensure the line `new file mode 100644` is included to specify file permissions.
 5. When deleting a file, include `deleted file mode` to indicate that the file has been removed. Each line in the 
 deleted file should be prefixed with `-` to indicate the content removal.
+6. Lines ending with a newline (\n) should not be interpreted as an additional line. Treat \n as the end of the current 
+line’s content, not as a new line in the file.
 
 Do not include any explanatory text within the diff blocks. If you need to provide explanations or comments, do so outside the diff blocks.
 
 The codebase is provided at the end of this prompt in a specific format. 
 The code that the user has given to you for context is in the format like below where first line has the File path and then the content follows.
 Each file starts with "File: <filepath>" followed by its content on subsequent lines. 
-Remember, each line of the file content now starts with a line number, beginning from 1. Use these numbers to generate 
-accurate hunk diffs, but do not include the line numbers in the actual diff content.
 
 File: <filepath>
 <Content of the file>. 

app/main.py

@@ -23,8 +23,6 @@ def parse_arguments():
                         help="Port number to run Ziya frontend on (e.g., --port 8080)")
     parser.add_argument("--version", action="store_true",
                         help="Prints the version of Ziya")
-    parser.add_argument("--enable-code-apply", action="store_true",
-                        help="Enable the feature to show the button to apply code diffs (unstable)")
     parser.add_argument("--max-depth", type=int, default=15,
                         help="Maximum depth for folder structure traversal (e.g., --max-depth 20)")
     return parser.parse_args()
@@ -40,9 +38,6 @@ def setup_environment(args):
         os.environ["ZIYA_AWS_PROFILE"] = args.profile
     if args.model:
         os.environ["ZIYA_AWS_MODEL"] = args.model
-    if args.enable_code_apply:
-        os.environ["ZIYA_ENABLE_CODE_APPLY"] = "true"
-
     os.environ["ZIYA_MAX_DEPTH"] = str(args.max_depth)
 
 

app/server.py

@@ -14,7 +14,7 @@ from pydantic import BaseModel
 # import pydevd_pycharm
 import uvicorn
 
-from app.utils.code_util import us_git_to_apply_code_diff, correct_git_diff
+from app.utils.code_util import use_git_to_apply_code_diff, correct_git_diff
 from app.utils.directory_util import get_ignored_patterns
 from app.utils.logging_utils import logger
 from app.utils.gitignore_parser import parse_gitignore_patterns
@@ -38,12 +38,8 @@ add_routes(app, agent_executor, disabled_endpoints=["playground"], path="/ziya")
 
 @app.get("/")
 async def root(request: Request):
-    enable_code_apply = os.environ.get("ZIYA_ENABLE_CODE_APPLY", "false")
-    enable_code_apply_bool = enable_code_apply.lower() == "true"
-    logger.info(f"enable_code_apply_bool: {enable_code_apply_bool}")
     return templates.TemplateResponse("index.html", {
-        "request": request,
-        "enable_code_apply": str(enable_code_apply_bool).lower()
+        "request": request
     })
 
 
@@ -120,8 +116,10 @@ async def apply_changes(request: ApplyChangesRequest):
         if not user_codebase_dir:
             raise ValueError("ZIYA_USER_CODEBASE_DIR environment variable is not set")
 
-        corrected_diff = correct_git_diff(request.diff)
-        us_git_to_apply_code_diff(corrected_diff)
+        file_path = os.path.join(user_codebase_dir, request.filePath)
+        corrected_diff = correct_git_diff(request.diff, file_path)
+        logger.info(f"corrected diff content: \n{corrected_diff}")
+        use_git_to_apply_code_diff(corrected_diff)
         return {'message': 'Changes applied successfully'}
     except Exception as e:
         logger.error(f"Error applying changes: {str(e)}", exc_info=True)

app/utils/code_util.py

@@ -6,7 +6,7 @@ import re
 
 HUNK_HEADER_REGEX = re.compile(r'^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@')
 
-def us_git_to_apply_code_diff(git_diff: str):
+def use_git_to_apply_code_diff(git_diff: str):
     """
     Apply a git diff to the user's codebase.
 
@@ -53,7 +53,7 @@ def us_git_to_apply_code_diff(git_diff: str):
     finally:
         os.remove(temp_file)
 
-def correct_git_diff(git_diff: str) -> str:
+def correct_git_diff(git_diff: str, original_file_path: str) -> str:
     """
     Corrects the hunk headers in a git diff string by recalculating the line counts and
     adjusting the starting line numbers in the new file, considering the cumulative effect of
@@ -64,6 +64,7 @@ def correct_git_diff(git_diff: str) -> str:
 
     Parameters:
         git_diff (str): The git diff string to be corrected. It may contain multiple hunks
+        original_file_path (str): Path to the original file to calculate correct start_line_old
                         and incorrect hunk headers.
 
     Returns:
@@ -83,6 +84,20 @@ def correct_git_diff(git_diff: str) -> str:
     # Split the diff into lines
     lines = git_diff.split('\n')
 
+    # Check if this is a new file creation by looking for "new file mode" in the diff
+    is_new_file = any('new file mode 100644' in line for line in lines[:5])
+    original_content = []
+
+    if not is_new_file:
+        try:
+            with open(original_file_path, 'r') as f:
+                original_content = f.read().splitlines()
+        except FileNotFoundError:
+            error_msg = (
+                f"File {original_file_path} not found and diff does not indicate new file creation. "
+            )
+            raise FileNotFoundError(error_msg)
+
     corrected_lines = []
     line_index = 0
 
@@ -96,8 +111,8 @@ def correct_git_diff(git_diff: str) -> str:
 
         if hunk_match:
             # Process the hunk
-            corrected_hunk_header, hunk_lines, line_index, line_offset = _process_hunk(
-                lines, line_index, cumulative_line_offset
+            corrected_hunk_header, hunk_lines, line_index, line_offset = _process_hunk_with_original_content(
+                lines, line_index, cumulative_line_offset, original_content
             )
 
             # Update cumulative_line_offset
@@ -115,7 +130,74 @@ def correct_git_diff(git_diff: str) -> str:
     corrected_diff = '\n'.join(corrected_lines)
     return corrected_diff
 
-def _process_hunk(lines: list, start_index: int, cumulative_line_offset: int):
+def _find_correct_old_start_line(original_content: list, hunk_lines: list) -> int:
+    """
+    Finds the correct starting line number in the original file by matching context and deleted lines.
+
+    Parameters:
+        original_content (list): List of lines from the original file
+        hunk_lines (list): List of lines in the current hunk
+
+    Returns:
+        int: The correct 1-based line number where the hunk should start in the original file
+
+    The function works by:
+    1. Extracting context and deleted lines from the hunk (ignoring added lines)
+    2. Creating a pattern from these lines
+    3. Finding where this pattern matches in the original file
+    4. Converting the matching position to a 1-based line number
+    """
+    # Extract context and deleted lines from the hunk
+    if not original_content:
+        # Creating a new file, should start with @@ -0,0 +1,N @@
+        return 0
+
+    if len(hunk_lines) < 3:
+        error_msg = (
+            f"Invalid git diff format: Expected at least 2 lines in the hunk, but got {len(hunk_lines)} lines.\n"
+            + "Hunk content:\n{}".format('\n'.join(hunk_lines)))
+        logger.error(error_msg)
+        raise RuntimeError("Invalid git diff format.")
+
+    context_and_deleted = []
+    for line in hunk_lines:
+        if line.startswith(' ') or line.startswith('-'):
+            # Remove the prefix character
+            context_and_deleted.append(line[1:])
+
+    if not context_and_deleted:
+        error_msg = (
+            "Invalid git diff format: No context or deleted lines found in the hunk.\n"
+            "Each hunk must contain at least one context line (starting with space) "
+            "or deleted line (starting with '-').\n"
+            "Hunk content:\n{}".format('\n'.join(hunk_lines)))
+        raise RuntimeError(error_msg)
+
+    # Search for the pattern in the original file
+    pattern_length = len(context_and_deleted)
+    for i in range(len(original_content) - pattern_length + 1):
+        matches = True
+        for j in range(pattern_length):
+            if j >= len(context_and_deleted):
+                break
+            if i + j >= len(original_content) or original_content[i + j] != context_and_deleted[j]:
+                matches = False
+                break
+        if matches:
+            # Found the correct position git diff start with 1.
+            return i + 1
+
+    joined_context_and_deleted = '\n'.join(context_and_deleted)
+    error_msg = (
+        "Failed to locate the hunk position in the original file.\n"
+        "This usually happens when the context lines in the diff don't match the original file content.\n"
+        f"Context and deleted lines being searched:\n{joined_context_and_deleted}\n"
+        "Please ensure the diff is generated against the correct version of the file."
+    )
+    logger.error(error_msg)
+    raise RuntimeError(error_msg)
+
+def _process_hunk_with_original_content(lines: list, start_index: int, cumulative_line_offset: int, original_content: list):
     """
     Processes a single hunk starting at start_index in lines, recalculates the line counts,
     and returns the corrected hunk header, hunk lines, and the updated index after the hunk.
@@ -124,6 +206,7 @@ def _process_hunk(lines: list, start_index: int, cumulative_line_offset: int):
         lines (list): The list of lines from the diff.
         start_index (int): The index in lines where the hunk header is located.
         cumulative_line_offset (int): The cumulative line offset from previous hunks.
+        original_content (list): List of lines from the original file.
 
     Returns:
         tuple:
@@ -137,11 +220,6 @@ def _process_hunk(lines: list, start_index: int, cumulative_line_offset: int):
     """
 
     line_index = start_index
-    hunk_header_line = lines[line_index]
-    hunk_match = HUNK_HEADER_REGEX.match(hunk_header_line)
-
-    # Extract the starting line numbers from the hunk header
-    start_line_old = int(hunk_match.group(1))
 
     # Initialize counts for recalculation
     actual_count_old = 0
@@ -161,7 +239,10 @@ def _process_hunk(lines: list, start_index: int, cumulative_line_offset: int):
             hunk_lines.append(hunk_line)
             line_index += 1
 
-    # Now process hunk_lines to calculate counts
+    # Find the correct start_line_old by matching context and deleted lines
+    start_line_old = _find_correct_old_start_line(original_content, hunk_lines)
+
+    # Calculate counts for the hunk lines
     for hunk_line in hunk_lines:
         if hunk_line.startswith('+') and not hunk_line.startswith('+++'):
             actual_count_new += 1
@@ -172,8 +253,15 @@ def _process_hunk(lines: list, start_index: int, cumulative_line_offset: int):
             actual_count_old += 1
             actual_count_new += 1
 
-    # Adjust start_line_new considering previous line offsets
-    corrected_start_line_new = start_line_old + cumulative_line_offset
+    # Special handling for new file creation
+    if start_line_old == 0:
+        # For new files:
+        # count_old should be 0
+        actual_count_old = 0
+        corrected_start_line_new = 1
+    else:
+        # For existing files, adjust start_line_new considering previous line offsets
+        corrected_start_line_new = start_line_old + cumulative_line_offset
 
     # Calculate line offset for subsequent hunks
     line_offset = actual_count_new - actual_count_old
@@ -211,24 +299,3 @@ def _format_hunk_header(start_old: int, count_old: int, start_new: int, count_ne
     if count_new != 1:
         new_part += f',{count_new}'
     return f'@@ {old_part} {new_part} @@'
-
-
-
-if __name__ == '__main__':
-    # TODO: Create unit test and move these code to unit test
-    diff = """\
-diff --git a/file.txt b/file.txt
-index e69de29..4b825dc 100644
---- a/file.txt
-+++ b/file.txt
-@@ -1,5 +1,5 @@
- Line one
-+Line two
- Line three
- Line four
- Line five
-@@ -10,2 +10,2 @@
- Line ten
--Line eleven
-+Line eleven modified"""
-    print(correct_git_diff(diff))

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ziya"
-version = "0.1.42"
+version = "0.1.44"
 description = ""
 authors = ["Vishnu Krishnaprasad <vishnukool@gmail.com>"]
 readme = "README.md"
@@ -25,6 +25,9 @@ langchain-cli = ">=0.0.15"
 pydevd-pycharm = "^243.18137.19"
 langchain-community = "^0.3.1"
 
+[tool.poetry.group.dev.dependencies]
+pytest = "^8.3.3"
+
 [build-system]
 requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"

templates/asset-manifest.json

@@ -1,7 +1,7 @@
 {
   "files": {
     "main.css": "/static/css/main.8af23da0.css",
-    "main.js": "/static/js/main.fee8aad7.js",
+    "main.js": "/static/js/main.50c95184.js",
     "static/media/fa-solid-900.ttf": "/static/media/fa-solid-900.bacd5de623fb563b961a.ttf",
     "static/media/fa-brands-400.ttf": "/static/media/fa-brands-400.60127e352b7a11f7f1bc.ttf",
     "static/media/fa-solid-900.woff2": "/static/media/fa-solid-900.4d986b00ff9ca3828fbd.woff2",
@@ -12,10 +12,10 @@
     "static/media/fa-v4compatibility.woff2": "/static/media/fa-v4compatibility.cf7f5903d06b79ad60f1.woff2",
     "index.html": "/index.html",
     "main.8af23da0.css.map": "/static/css/main.8af23da0.css.map",
-    "main.fee8aad7.js.map": "/static/js/main.fee8aad7.js.map"
+    "main.50c95184.js.map": "/static/js/main.50c95184.js.map"
   },
   "entrypoints": [
     "static/css/main.8af23da0.css",
-    "static/js/main.fee8aad7.js"
+    "static/js/main.50c95184.js"
   ]
 }

templates/index.html

@@ -1 +1 @@
-<!doctype html><html lang="en"><head><meta charset="UTF-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>Ziya - Code Assistant</title><link rel="icon" href="/favicon.ico" type="image/x-icon"><script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script><script>window.enableCodeApply="{{ enable_code_apply | lower }}"</script><script defer="defer" src="/static/js/main.fee8aad7.js"></script><link href="/static/css/main.8af23da0.css" rel="stylesheet"></head><body><div id="root"></div></body></html>
+<!doctype html><html lang="en"><head><meta charset="UTF-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>Ziya - Code Assistant</title><link rel="icon" href="/favicon.ico" type="image/x-icon"><script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script><script>window.enableCodeApply="true"</script><script defer="defer" src="/static/js/main.50c95184.js"></script><link href="/static/css/main.8af23da0.css" rel="stylesheet"></head><body><div id="root"></div></body></html>