check point

2025-12-25 18:06:29 +07:00
commit 31de8b0d84
34 changed files with 3209 additions and 0 deletions
--- a/backup_source/match.py
+++ b/backup_source/match.py
@@ -0,0 +1,81 @@
+"""
+games/match.py - Match Game - Match sentences with images
+"""
+from typing import List
+from pydantic import BaseModel, Field
+from langchain_core.output_parsers import PydanticOutputParser
+
+
+# ============== SCHEMA ==============
+class MatchItem(BaseModel):
+    word: str = Field(description="The sentence to be matched (EXACT copy from source)")
+    match_with: str = Field(description="Short keyword for reference")
+    original_quote: str = Field(description="EXACT quote from source text")
+    image_description: str = Field(default="", description="Detailed visual description for image generation/search")
+    image_is_complex: bool = Field(default=False, description="True if image needs precise quantities, humans, or multiple detailed objects")
+
+
+class MatchOutput(BaseModel):
+    """Output wrapper for match items"""
+    items: List[MatchItem] = Field(description="List of match items generated from source text")
+
+
+# Output parser
+output_parser = PydanticOutputParser(pydantic_object=MatchOutput)
+
+
+# ============== CONFIG ==============
+GAME_CONFIG = {
+    "game_type": "match",
+    "display_name": "Match with Image",
+    "description": "Match sentences with images",
+    
+    "active": True,
+    
+    "min_items": 2,
+    "max_items": 10,
+    "schema": MatchItem,
+    "output_schema": MatchOutput,
+    "output_parser": output_parser,
+    
+    "system_prompt": """Extract sentences and create image descriptions for matching game.
+The game will show images and players must match them with the correct sentences.
+
+YOUR TASK:
+1. Extract meaningful sentences from the source text
+2. Create a DETAILED image_description that clearly represents the sentence
+3. The image should be distinct enough to match with its sentence
+
+CRITICAL RULES:
+1. KEEP THE ORIGINAL LANGUAGE - Do NOT translate the source text
+2. original_quote MUST be an EXACT copy from source text
+3. image_description must be DETAILED and SPECIFIC to the sentence content
+4. Each image should be visually distinguishable from others""",
+}
+
+
+# ============== EXAMPLES ==============
+EXAMPLES = [
+    {
+        "input": "The Sun is a star. The Moon orbits Earth.",
+        "output": {
+            "items": [
+                {
+                    "word": "The Sun is a star.",
+                    "match_with": "sun",
+                    "original_quote": "The Sun is a star.",
+                    "image_description": "A bright glowing yellow sun with solar flares",
+                    "image_is_complex": False
+                },
+                {
+                    "word": "The Moon orbits Earth.",
+                    "match_with": "moon",
+                    "original_quote": "The Moon orbits Earth.",
+                    "image_description": "A grey moon circling around the blue Earth planet",
+                    "image_is_complex": False
+                }
+            ]
+        },
+        "why_suitable": "Has distinct concepts that can be visualized and matched"
+    }
+]
--- a/backup_source/memory_card.py
+++ b/backup_source/memory_card.py
@@ -0,0 +1,61 @@
+"""
+games/memory_card.py - Memory Card Game - Flip cards to find pairs
+"""
+from typing import List
+from pydantic import BaseModel, Field
+from langchain_core.output_parsers import PydanticOutputParser
+
+
+# ============== SCHEMA ==============
+class MemoryCardItem(BaseModel):
+    name: str = Field(description="Card content/label")
+    pair_id: str = Field(description="ID to match pairs (same pair_id = matching cards)")
+    original_quote: str = Field(description="EXACT quote from source text")
+    image_description: str = Field(default="", description="Visual description for the card")
+    image_is_complex: bool = Field(default=False, description="True if image needs precise quantities, humans, or multiple detailed objects")
+
+
+class MemoryCardOutput(BaseModel):
+    """Output wrapper for memory card items"""
+    items: List[MemoryCardItem] = Field(description="List of memory card items generated from source text")
+
+
+# Output parser
+output_parser = PydanticOutputParser(pydantic_object=MemoryCardOutput)
+
+
+# ============== CONFIG ==============
+GAME_CONFIG = {
+    "game_type": "memory_card",
+    "display_name": "Memory Card",
+    "description": "Flip cards to find pairs",
+    
+    "active": False,  # Disabled
+    
+    "min_items": 4,
+    "max_items": 10,
+    "schema": MemoryCardItem,
+    "output_schema": MemoryCardOutput,
+    "output_parser": output_parser,
+    
+    "system_prompt": """Create memory card pairs.
+CRITICAL RULES:
+1. KEEP THE ORIGINAL LANGUAGE - Do NOT translate the source text
+2. original_quote MUST be an EXACT copy from source text
+3. ALL content must come from the source text only""",
+}
+
+
+# ============== EXAMPLES ==============
+EXAMPLES = [
+    {
+        "input": "The Sun is a star.",
+        "output": {
+            "items": [
+                {"name": "The Sun", "pair_id": "p1", "original_quote": "The Sun is a star.", "image_description": "A bright sun", "image_is_complex": False},
+                {"name": "a star", "pair_id": "p1", "original_quote": "The Sun is a star.", "image_description": "A glowing star", "image_is_complex": False}
+            ]
+        },
+        "why_suitable": "Has concept pairs"
+    }
+]
--- a/backup_source/sequence_sentence.py
+++ b/backup_source/sequence_sentence.py
@@ -0,0 +1,127 @@
+"""
+games/sequence_sentence.py - Arrange Sentences Game
+type_id = 2
+"""
+from typing import List
+from pydantic import BaseModel, Field
+from langchain_core.output_parsers import PydanticOutputParser
+
+
+# ============== SCHEMA ==============
+class SentenceItem(BaseModel):
+    sentence: str = Field(description="Full sentence to arrange (EXACT from source)")
+    original_quote: str = Field(description="EXACT quote from source text")
+    image_description: str = Field(default="", description="Visual description of the content")
+    image_keywords: List[str] = Field(default=[], description="Keywords for image search")
+    image_is_complex: bool = Field(default=False, description="True if image needs precise quantities, humans, or multiple detailed objects")
+
+
+class SentenceMetadata(BaseModel):
+    """Metadata đánh giá nội dung"""
+    title: str = Field(
+        description="Title for this content. Prefer title from source document if available and suitable, otherwise create a short descriptive title."
+    )
+    description: str = Field(
+        description="Short description summarizing the content/topic."
+    )
+    grade: int = Field(
+        description="Estimated grade level 1-5 (1=easy/young, 5=advanced/older). Judge by vocabulary, concepts."
+    )
+    type: str = Field(default="sequence_sentence", description="Game type")
+    difficulty: int = Field(
+        description="Difficulty 1-5 for that grade (1=very easy, 5=very hard). Judge by sentence complexity, vocabulary."
+    )
+
+
+class SentenceOutput(BaseModel):
+    """Output wrapper for sentence items"""
+    items: List[SentenceItem] = Field(description="List of sentence items generated from source text")
+    metadata: SentenceMetadata = Field(description="Metadata about the content")
+
+
+# Output parser
+output_parser = PydanticOutputParser(pydantic_object=SentenceOutput)
+
+
+# ============== CONFIG ==============
+GAME_CONFIG = {
+    "game_type": "sequence_sentence",
+    "display_name": "Arrange Sentences",
+    "description": "Arrange sentences in order",
+    "type_id": 2,
+    
+    "active": True,
+    
+    "max_items": 10,
+    "schema": SentenceItem,
+    "output_schema": SentenceOutput,
+    "output_parser": output_parser,
+    
+    # Dùng cho analyze + generate (không có format rules)
+    "system_prompt": """Extract sentences from source text.
+
+RULES:
+1. KEEP THE ORIGINAL LANGUAGE - Do NOT translate
+2. sentence = EXACT copy from source text
+3. original_quote = same as sentence value
+4. image_description = ALWAYS provide a short visual description (NEVER empty)
+5. image_is_complex = FALSE for simple/static objects, TRUE for quantities/humans/complex scenes""",
+    
+    # Dùng cho generate trực tiếp (CÓ format rules)
+    "direct_prompt": """Extract sentences from source text.
+
+EXPECTED INPUT: List of sentences (separated by semicolon, newline, or similar)
+
+STEP 1 - VALIDATE INPUT:
+Analyze if input looks like a list of sentences suitable for "arrange sentences" game.
+- Should contain multiple complete sentences
+- Should NOT be a quiz, single word list, or Q&A format
+
+If input is clearly NOT suitable (e.g. it's a quiz, single words only, or wrong format), return:
+{{"items": [], "format_error": "Input không phù hợp cho game sắp xếp câu"}}
+
+STEP 2 - EXTRACT (if valid):
+RULES:
+1. KEEP THE ORIGINAL LANGUAGE - Do NOT translate
+2. Extract ALL sentences from source
+3. sentence = EXACT sentence from source (trim whitespace)
+4. original_quote = same as sentence value
+5. image_description = ALWAYS provide a short visual description (NEVER leave empty)
+6. image_is_complex:
+   - FALSE: simple objects, static things, no specific quantities (e.g. "a sun", "a tree")
+   - TRUE: needs exact quantities, humans/people, or complex details (e.g. "3 birds", "a boy reading")""",
+}
+
+
+# ============== EXAMPLES ==============
+EXAMPLES = [
+    {
+        "input": "The Sun is a star; The Moon orbits Earth; Mars is red",
+        "output": {
+            "items": [
+                {
+                    "sentence": "The Sun is a star",
+                    "original_quote": "The Sun is a star",
+                    "image_description": "A bright glowing sun",
+                    "image_keywords": ["sun", "star"],
+                    "image_is_complex": False
+                },
+                {
+                    "sentence": "The Moon orbits Earth",
+                    "original_quote": "The Moon orbits Earth",
+                    "image_description": "Moon circling around Earth",
+                    "image_keywords": ["moon", "earth", "orbit"],
+                    "image_is_complex": False
+                },
+                {
+                    "sentence": "Mars is red",
+                    "original_quote": "Mars is red",
+                    "image_description": "Red planet Mars",
+                    "image_keywords": ["mars", "red", "planet"],
+                    "image_is_complex": False
+                }
+            ]
+        },
+        "why_suitable": "Source has sentences separated by semicolons"
+    }
+]
--- a/backup_source/sequence_word.py
+++ b/backup_source/sequence_word.py
@@ -0,0 +1,134 @@
+"""
+games/sequence_word.py - Arrange Words Game
+type_id = 3
+"""
+from typing import List
+from pydantic import BaseModel, Field
+from langchain_core.output_parsers import PydanticOutputParser
+
+
+# ============== SCHEMA ==============
+class WordItem(BaseModel):
+    word: str = Field(description="Word or phrase to arrange (EXACT from source)")
+    original_quote: str = Field(description="EXACT quote from source text")
+    image_description: str = Field(default="", description="Visual description of the content")
+    image_keywords: List[str] = Field(default=[], description="Keywords for image search")
+    image_is_complex: bool = Field(default=False, description="True if image needs precise quantities, humans, or multiple detailed objects")
+
+
+class WordMetadata(BaseModel):
+    """Metadata đánh giá nội dung"""
+    title: str = Field(
+        description="Title for this content. Prefer title from source document if available and suitable, otherwise create a short descriptive title."
+    )
+    description: str = Field(
+        description="Short description summarizing the content/topic."
+    )
+    grade: int = Field(
+        description="Estimated grade level 1-5 (1=easy/young, 5=advanced/older). Judge by vocabulary complexity."
+    )
+    type: str = Field(default="sequence_word", description="Game type")
+    difficulty: int = Field(
+        description="Difficulty 1-5 for that grade (1=very easy, 5=very hard). Judge by word complexity, number of items."
+    )
+
+
+class WordOutput(BaseModel):
+    """Output wrapper for word items"""
+    items: List[WordItem] = Field(description="List of word items generated from source text")
+    metadata: WordMetadata = Field(description="Metadata about the content")
+
+
+# Output parser
+output_parser = PydanticOutputParser(pydantic_object=WordOutput)
+
+
+# ============== CONFIG ==============
+GAME_CONFIG = {
+    "game_type": "sequence_word",
+    "display_name": "Arrange Words",
+    "description": "Arrange words or phrases in order",
+    "type_id": 3,
+    
+    "active": True,
+    
+    "max_items": 10,
+    "schema": WordItem,
+    "output_schema": WordOutput,
+    "output_parser": output_parser,
+    
+    # Dùng cho analyze + generate (không có format rules)
+    "system_prompt": """Extract words or phrases from source text.
+
+RULES:
+1. KEEP THE ORIGINAL LANGUAGE - Do NOT translate
+2. word = EXACT copy from source text
+3. original_quote = same as word value
+4. image_description = ALWAYS provide a short visual description (NEVER empty)
+5. image_is_complex = FALSE for simple/static objects, TRUE for quantities/humans/complex scenes""",
+    
+    # Dùng cho generate trực tiếp (CÓ format rules)
+    "direct_prompt": """Extract words or phrases from source text.
+
+EXPECTED INPUT: List of words/phrases (separated by semicolon, comma, newline, or similar)
+
+STEP 1 - VALIDATE INPUT:
+Analyze if input looks like a list of words/phrases suitable for "arrange words" game.
+- Should contain multiple short words or phrases
+- Should NOT be a paragraph, essay, or Q&A format
+
+If input is clearly NOT suitable (e.g. it's a quiz, a long paragraph, or wrong format), return:
+{{"items": [], "format_error": "Input không phù hợp cho game sắp xếp từ"}}
+
+STEP 2 - EXTRACT (if valid):
+RULES:
+1. KEEP THE ORIGINAL LANGUAGE - Do NOT translate
+2. Extract ALL words/phrases from source
+3. word = EXACT word/phrase from source (trim whitespace)
+4. original_quote = same as word value
+5. image_description = ALWAYS provide a short visual description (NEVER leave empty)
+6. image_is_complex:
+   - FALSE: simple objects, static things, no specific quantities (e.g. "an apple", "a book")
+   - TRUE: needs exact quantities, humans/people, or complex details (e.g. "5 oranges", "a woman cooking")""",
+}
+
+
+# ============== EXAMPLES ==============
+EXAMPLES = [
+    {
+        "input": "Apple; Banana; Orange; Grape",
+        "output": {
+            "items": [
+                {
+                    "word": "Apple",
+                    "original_quote": "Apple",
+                    "image_description": "A red apple",
+                    "image_keywords": ["apple"],
+                    "image_is_complex": False
+                },
+                {
+                    "word": "Banana",
+                    "original_quote": "Banana",
+                    "image_description": "A yellow banana",
+                    "image_keywords": ["banana"],
+                    "image_is_complex": False
+                },
+                {
+                    "word": "Orange",
+                    "original_quote": "Orange",
+                    "image_description": "An orange fruit",
+                    "image_keywords": ["orange"],
+                    "image_is_complex": False
+                },
+                {
+                    "word": "Grape",
+                    "original_quote": "Grape",
+                    "image_description": "Purple grapes",
+                    "image_keywords": ["grape"],
+                    "image_is_complex": False
+                }
+            ]
+        },
+        "why_suitable": "Source has words separated by semicolons"
+    }
+]