syntaxmatrix 2.2.8__tar.gz → 2.3.0__tar.gz

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: syntaxmatrix
-Version: 2.2.8
+Version: 2.3.0
 Summary: SyntaxMUI: A customizable framework for Python AI Assistant Projects.
 Home-page: https://github.com/bobganti/syntaxmatrix_demo
 Author: Bob Nti
@@ -24,6 +24,7 @@ Requires-Dist: openai>=1.84.0
 Requires-Dist: google-genai>=1.19.0
 Requires-Dist: anthropic>=0.67.0
 Requires-Dist: reportlab>=4.4.3
+Requires-Dist: lxml>=6.0.2
 Provides-Extra: mlearning
 Requires-Dist: pandas>=2.2.3; extra == "mlearning"
 Requires-Dist: numpy>=2.0.2; extra == "mlearning"

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/SyntaxMatrix.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: syntaxmatrix
-Version: 2.2.8
+Version: 2.3.0
 Summary: SyntaxMUI: A customizable framework for Python AI Assistant Projects.
 Home-page: https://github.com/bobganti/syntaxmatrix_demo
 Author: Bob Nti
@@ -24,6 +24,7 @@ Requires-Dist: openai>=1.84.0
 Requires-Dist: google-genai>=1.19.0
 Requires-Dist: anthropic>=0.67.0
 Requires-Dist: reportlab>=4.4.3
+Requires-Dist: lxml>=6.0.2
 Provides-Extra: mlearning
 Requires-Dist: pandas>=2.2.3; extra == "mlearning"
 Requires-Dist: numpy>=2.0.2; extra == "mlearning"

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/SyntaxMatrix.egg-info/requires.txt

@@ -10,6 +10,7 @@ openai>=1.84.0
 google-genai>=1.19.0
 anthropic>=0.67.0
 reportlab>=4.4.3
+lxml>=6.0.2
 
 [:sys_platform == "win32"]
 pywin32>=311

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/setup.py

@@ -8,7 +8,7 @@ with open(os.path.join(this_directory, "README.md"), encoding="utf-8") as f:
 
 setup(
     name="syntaxmatrix",
-    version="2.2.8",
+    version="2.3.0",
     author="Bob Nti",
     author_email="bob.nti@syntaxmatrix.com",
     description="SyntaxMUI: A customizable framework for Python AI Assistant Projects.",
@@ -38,6 +38,7 @@ setup(
         "google-genai>=1.19.0",
         "anthropic>=0.67.0",
         "reportlab>=4.4.3",
+        "lxml>=6.0.2"
     ],
     extras_require={
         "mlearning": [

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/syntaxmatrix/__init__.py

@@ -59,6 +59,6 @@ stream_write = _app_instance.stream_write
 enable_stream = _app_instance.enable_stream
 stream = _app_instance.stream
 get_stream_args = _app_instance.get_stream_args
-# set_stream_args = _app_instance.set_stream_args
 
-app = _app_instance.app
+
+app = _app_instance.app

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/syntaxmatrix/commentary.py

@@ -153,35 +153,33 @@ def _context_strings(context: Dict[str, Any]) -> List[str]:
     return out
 
 
-# ---LLM prompts ---
-
-_SYSTEM_VISION = (
+def phrase_commentary_vision(context: Dict[str, Any], images_b64: List[str]) -> str:
+    """
+    Use the project's 'vision2text' profile (profiles.py). If the provider supports images,
+    send figures + text; otherwise fall back to a text-only prompt grounded by labels.
+    """
+    
+    _SYSTEM_VISION = (
     "You are a data analyst. Write a short, precise commentary in UK English that explains what the "
     "already-rendered visuals mean for the user's question. "
     "Use information visible in the attached figures and the provided context strings (field names, labels). "
     "Do not invent numbers. If the figures/context are insufficient, say: 'Insufficient context to comment usefully.'"
-)
+    )
 
-_USER_TMPL_VISION = """\
-Question:
-{q}
+    _USER_TMPL_VISION = """\
+    question:
+    {q}
 
-Visible context strings (titles, axes, legends, headers):
-{ctx}
+    Visible context strings (titles, axes, legends, headers):
+    {ctx}
 
-Write a concise conclusion (~200-260 words) with:
-- <strong>Headline</strong> (one sentence answering the question).
-- <strong>Evidence</strong> (6-8 bullets referencing panels/axes/legend groups seen in the figures and explaining the plots/tables  vis-a-vis the query). Explain all the oupupt comprehensively in details.
-- <strong>Limitations</strong> (1 bullets; avoid quoting numbers unless present in context).
-- <strong>Next step</strong> (1 bullet).
-"""
-
-def phrase_commentary_vision(context: Dict[str, Any], images_b64: List[str]) -> str:
+    Write a concise conclusion (~200-260 words) with:
+    - <strong>Headline</strong> (one sentence answering the question).
+    - <strong>Evidence</strong> (6-8 bullets referencing panels/axes/legend groups seen in the figures and explaining the plots/tables  vis-a-vis the query). Explain all the oupupt comprehensively in details.
+    - <strong>Limitations</strong> (1 bullets; avoid quoting numbers unless present in context).
+    - <strong>Next step</strong> (1 bullet).
     """
-    Use the project's 'vision2text' profile (profiles.py). If the provider supports images,
-    send figures + text; otherwise fall back to a text-only prompt grounded by labels.
-    """
-  
+
     visible = _context_strings(context)
     user = _USER_TMPL_VISION.format(
         q=context.get("question",""),
@@ -202,22 +200,36 @@ def phrase_commentary_vision(context: Dict[str, Any], images_b64: List[str]) ->
     # Google
     if _provider == "google":
         try:
-            parts = [types.Part.from_text(text=user)]
+            # Gemini expects a different structure
+            contents = []
+            
+            # Add text part first
+            text_part = {
+                "text": _SYSTEM_VISION + "\n\n" + user
+            }
+            contents.append(text_part)
+            
+            # Add image parts
             for b64 in images_b64[:4]:
                 if b64:
-                    parts.append(types.Part.from_bytes(data=base64.b64decode(b64), mime_type="image/png"))
-            contents = [types.Content(role="user", parts=parts)]
-            resp = _client.models.generate_content(
+                    image_part = {
+                        "inline_data": {
+                            "mime_type": "image/png",
+                            "data": b64
+                        }
+                    }
+                    contents.append(image_part)
+            
+            # Correct Gemini API call
+            response = _client.models.generate_content(
                 model=_model,
-                contents=contents,
-                system_instruction=_SYSTEM_VISION
+                contents=contents
             )
-            txt = getattr(resp, "text", "") or ""
-            if txt.strip():
-                return txt.strip()
-        except Exception:
-             pass  # Fall through to the chat.completions fallback implemented below.
-    
+            txt = response.text.strip()
+            return txt.strip()
+        except Exception as e:
+            return f"Google Gemini error: {e}"
+           
     # Openai
     elif _provider == "openai" and _model in GPT_MODELS_LATEST:
         # Use the Responses API with multimodal input (text + up to 4 images)

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/syntaxmatrix/core.py

@@ -14,7 +14,7 @@ from .file_processor import process_admin_pdf_files
 from google.genai import types
 from .vector_db import query_embeddings
 from .vectorizer import embed_text
-from syntaxmatrix.settings.prompts import SMXAI_CHAT_ID, SMXAI_CHAT_INSTRUCTIONS, SMX_WEBSITE_DESCRIPTION
+from syntaxmatrix.settings.prompts import SMXAI_CHAT_ID, SMXAI_CHAT_INSTRUCTIONS, SMXAI_WEBSITE_DESCRIPTION
 from typing import List, Generator
 from .auth import init_auth_db
 from . import profiles as _prof
@@ -26,6 +26,7 @@ from dotenv import load_dotenv
 from html import unescape
 
 
+
 # ──────── framework‐local storage paths ────────
 # this ensures the key & data always live under the package dir,
 # regardless of where the developer `cd` into before launching.
@@ -71,7 +72,7 @@ class SyntaxMUI:
         self.user_files_enabled = False
         self.ai_chat_id = SMXAI_CHAT_ID
         self.ai_chat_instructions = SMXAI_CHAT_INSTRUCTIONS    
-        self.website_description = SMX_WEBSITE_DESCRIPTION
+        self.website_description = SMXAI_WEBSITE_DESCRIPTION
 
         db.init_db()
         self.page = ""
@@ -611,9 +612,12 @@ class SyntaxMUI:
         if _provider == "google":
             intent = google_classify_query()
             return intent
-        elif _model in self.gpt_models_latest():
+        if _model in self.gpt_models_latest():
             intent = gpt_models_latest_classify_query()
             return intent
+        if _provider == "anthropic":
+            intent = anthropic_classify_query()
+            return intent
         else:
             intent = openai_sdk_classify_query()
             return intent
@@ -811,7 +815,6 @@ class SyntaxMUI:
                         yield token
         except Exception as e:
             yield f"Error during streaming: {type(e).__name__}: {e}"
-
     
     def process_query(self, query, context, history, stream=False):
 
@@ -1005,13 +1008,83 @@ class SyntaxMUI:
         - Solves: {question},
         - And ends with at least one visible output (`show(...)` and/or `plt.show()`).
         """
-
+        
         def google_generate_code():
-            response = _client.models.generate_content(
-                model=_model, 
-                contents=f"{ai_profile}\n\n{instructions}"
-            )
-            return response.text
+            try:
+                # Combine system prompt and instructions for Gemini
+                full_prompt = f"{ai_profile}\n\n{instructions}"
+                
+                # Gemini expects a simple generate_content call with the model and contents
+                response = _client.models.generate_content(
+                    model=_model, 
+                    contents=full_prompt
+                )
+                
+                # Extract text from response
+                if hasattr(response, 'text'):
+                    return response.text
+                elif hasattr(response, 'candidates') and response.candidates:
+                    candidate = response.candidates[0]
+                    if hasattr(candidate.content, 'parts'):
+                        return ''.join(part.text for part in candidate.content.parts if hasattr(part, 'text'))
+                return str(response)
+                    
+            except Exception as e:
+                print(f"Google Gemini code generation error: {e}")
+                # Return a basic analysis code with ALL necessary imports
+                return """
+                import pandas as pd
+                import matplotlib.pyplot as plt
+                import seaborn as sns
+                import numpy as np
+                import io
+                import base64
+                from syntaxmatrix.display import show
+
+                print("Basic DataFrame Info:")
+                print(f"Shape: {df.shape}")
+                print("\\nColumns and dtypes:")
+                print(df.dtypes)
+                print("\\nBasic statistics:")
+                show(df.describe())
+
+                print("\\nFirst few rows:")
+                show(df.head())
+
+                # Generate a simple visualization based on available columns
+                plt.figure(figsize=(10, 6))
+
+                if len(df.columns) >= 2:
+                    # Try to find numeric columns for scatter plot
+                    numeric_cols = df.select_dtypes(include=['number']).columns
+                    if len(numeric_cols) >= 2:
+                        sns.scatterplot(data=df, x=numeric_cols[0], y=numeric_cols[1])
+                        plt.title(f"Scatter plot: {numeric_cols[0]} vs {numeric_cols[1]}")
+                        plt.tight_layout()
+                        plt.show()
+                    else:
+                        # Use first column for bar plot
+                        top_values = df[df.columns[0]].value_counts().head(10)
+                        top_values.plot(kind='bar')
+                        plt.title(f"Top 10 values in {df.columns[0]}")
+                        plt.tight_layout()
+                        plt.show()
+                else:
+                    # Single column analysis
+                    if len(df.columns) == 1:
+                        col_name = df.columns[0]
+                        if df[col_name].dtype in ['object', 'category']:
+                            df[col_name].value_counts().head(10).plot(kind='bar')
+                            plt.title(f"Value counts for {col_name}")
+                        else:
+                            df[col_name].hist(bins=20)
+                            plt.title(f"Distribution of {col_name}")
+                        plt.tight_layout()
+                        plt.show()
+                    else:
+                        print("Insufficient columns for detailed analysis")
+                        show(df)
+                """
 
         def gpt_models_latest_generate_code(reasoning_effort = "medium", verbosity = "medium"):
             try:                 
@@ -1077,6 +1150,24 @@ class SyntaxMUI:
 
             code = strip_describe_slice(code)
             code = drop_bad_classification_metrics(code, df)
+
+            if "import io" not in code and "io.BytesIO" in code:
+                lines = code.split('\n')
+                import_lines = []
+                other_lines = []
+                
+                for line in lines:
+                    if line.strip().startswith('import ') or line.strip().startswith('from '):
+                        import_lines.append(line)
+                    else:
+                        other_lines.append(line)
+                
+                # Add missing io import
+                if "import io" not in '\n'.join(import_lines):
+                    import_lines.append('import io')
+                
+                code = '\n'.join(import_lines + [''] + other_lines)
+        
             return code.strip()
     
 

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/syntaxmatrix/kernel_manager.py

@@ -48,7 +48,7 @@ class SyntaxMatrixKernelManager:
 
 _df_cache = None
 
-def execute_code_in_kernel(kc, code, timeout=90):
+def execute_code_in_kernel(kc, code, timeout=120):
 
     _local_stdout = ""
     _local_stderr = ""

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/syntaxmatrix/routes.py

@@ -50,9 +50,9 @@ _stream_q = queue.Queue()
 _stream_cancelled = {}
 _last_result_html = {}  # { session_id: html_doc }
 
-# store per-client so it stays alongside uploads
-MLEARNING_OUTPUTS = os.path.join(_CLIENT_DIR, "mlearning_outputs")
-os.makedirs(MLEARNING_OUTPUTS, exist_ok=True)
+# # store per-client so it stays alongside uploads
+# MLEARNING_OUTPUTS = os.path.join(_CLIENT_DIR, "mlearning_outputs")
+# os.makedirs(MLEARNING_OUTPUTS, exist_ok=True)
 
 # single, reused formatter: inline styles, padding, rounded corners, scroll
 _FMT = _HtmlFmt(
@@ -5230,7 +5230,7 @@ def setup_routes(smx):
                             )
                         ).strip()
                 
-            elif prof["provider"] == "openai":
+            elif prof["provider"] == "openai" and _model in smx.gpt_models_latest():
                 args = set_args(model=prof.get("model"), instructions=_CELL_REPAIR_RULES,
                           input=_prompt, previous_id=None, store=False,
                           reasoning_effort="minimal", verbosity="low")
@@ -5388,30 +5388,13 @@ def setup_routes(smx):
                     # 2.4 Append a single download button (explicit click → fetch → download)
                     download_url = url_for("download_result_html", session_id=session_id)
                     dl_html = f"""
-                      <button type="button"
-                              class="btn"
-                              style="margin:14px 0;padding:8px 12px;border:1px solid #0b6;border-radius:6px;background:#fff;color:#0b6;cursor:pointer;"
-                              onclick="(async function(btn) {{
-                                  try {{
-                                      btn.disabled = true;
-                                      const orig = btn.textContent;
-                                      btn.textContent = 'Saving…';
-                                      const resp = await fetch('{download_url}', {{
-                                          method: 'POST',
-                                          headers: {{ 'X-Requested-With': 'fetch' }}
-                                      }});
-                                      const data = await resp.json();
-                                      if (!resp.ok || !data.ok) throw new Error(data.error || 'Save failed');
-                                      btn.textContent = 'Saved to MLEARNING_OUTPUTS';
-                                  }} catch (err) {{
-                                      console.error(err);
-                                      alert('Could not save. Please try again.');
-                                  }} finally {{
-                                      btn.disabled = false;
-                                  }}
-                              }})(this)">
-                        Download
-                      </button>
+                      <a href="{download_url}">
+                        <button type="button"
+                                class="btn"
+                                style="margin:14px 0;padding:8px 12px;border:1px solid #0b6;border-radius:6px;background:#fff;color:#0b6;cursor:pointer;">
+                          Download
+                        </button>
+                      </a>
                     """
                     ai_outputs.append(Markup(dl_html))
 
@@ -5489,26 +5472,29 @@ def setup_routes(smx):
           data_cells=data_cells,
           session_id=session_id,
         )
+    
 
-
-    @smx.app.route("/download/result/html/<session_id>", methods=["POST"])
+    @smx.app.route("/download/result/html/<session_id>", methods=["GET"])
     def download_result_html(session_id):
-        """Send the last-built result HTML as a download (only when user clicks)."""
-        os.makedirs(MLEARNING_OUTPUTS, exist_ok=True)
-        stamp = datetime.now().strftime("%Y%m%d-%H%M%S-%f") 
-        filename = f"result_{session_id}_{stamp}.html"
-        path = os.path.join(MLEARNING_OUTPUTS, filename)
-
+        """Stream the last-built result HTML as a browser download (no server save)."""
         html_doc = _last_result_html.get(session_id)
         if not html_doc:
-            return jsonify({"ok": False, "error": "No result available."}), 404
-        with open(path, "w", encoding="utf-8") as f:
-            f.write(html_doc)
+            return ("No result available.", 404)
 
-        # free memory for this session copy (optional)
+        buf = io.BytesIO(html_doc.encode("utf-8"))
+        buf.seek(0)
+
+        # keep a copy if you wish, or free it:
         _last_result_html.pop(session_id, None)
 
-        return jsonify({"ok": True, "path": path})
+        stamp = datetime.now().strftime("%Y%m%d-%H%M%S-%f")
+        filename = f"result_{session_id}_{stamp}.html"
+        return send_file(
+            buf,
+            mimetype="text/html; charset=utf-8",
+            as_attachment=True,
+            download_name=filename
+        )
         
     # ── UPLOAD DATASET --------------------------------------
     @smx.app.route("/dashboard/upload", methods=["POST"])

{syntaxmatrix-2.2.8 → syntaxmatrix-2.3.0}/syntaxmatrix/settings/prompts.py

@@ -40,7 +40,7 @@ SMXAI_CHAT_INSTRUCTIONS = """
         8. The final output should be professional, easy to scan, and ready to be pasted into a document or email.
 """
 
-SMX_WEBSITE_DESCRIPTION = F"""
+SMXAI_WEBSITE_DESCRIPTION = F"""
     SyntaxMatrix Overview
     SyntaxMatrix is a battle-tested Python framework that accelerates AI application development from concept to production, slashing engineering overhead by up to 80%. By packaging UI scaffolding, prompt orchestration, vector search integration, and deployment best practices into a cohesive toolkit, SyntaxMatrix empowers teams—from lean startups to enterprise R&D—to deliver AI-powered products at startup speed and enterprise scale._
     ____________________________________
@@ -97,7 +97,7 @@ SMX_WEBSITE_DESCRIPTION = F"""
 """
 
 SMX_PAGE_GENERATION_INSTRUCTIONS = f"""
-    0· Parse the Website Description (MANDATORY):\n{SMX_WEBSITE_DESCRIPTION}\n\n
+    0· Parse the Website Description (MANDATORY):\n{SMXAI_WEBSITE_DESCRIPTION}\n\n
     1. Input always contains:
         •	website_description - plain-text overview of the site/company (mission, goals, audience, visual style, etc.).
         •	page_title - the specific page to create (e.g. About, Pricing, Blog).
@@ -163,7 +163,7 @@ SMX_PAGE_GENERATION_INSTRUCTIONS = f"""
 """
 
 WEBPAGE_GENERATION_INSTRUCTIONS = (f"""
-    Parse the Website Description (MANDATORY):\n{SMX_WEBSITE_DESCRIPTION}\n\n
+    Parse the Website Description (MANDATORY):\n{SMXAI_WEBSITE_DESCRIPTION}\n\n
     Prompt (give this to the LLM)
     Role: You are a senior front-end engineer and content designer building a single HTML page for the brand SyntaxMatrix. Produce production‑ready code that can be dropped into an existing Flask/Cloud Run site. Do not include a navbar or a footer; the host site renders them.