import os import json import argparse from openai import OpenAI if "OPENAI_API_KEY" not in os.environ: raise EnvironmentError("Missing OPENAI_API_KEY environment variable.") client = OpenAI(api_key=os.environ["OPENAI_API_KEY"]) # SCHEMAS must exist here or be imported # from schemas import SCHEMAS SCHEMAS = [ { "name": "Exclusivity Signal", "template": "I {exclusive_action} for {this_unique_thing}.", "description": "Signals commitment and self-signaling through decisive exclusivity.", "variables": { "exclusive_action": { "query": "What actions do people take, or what do they give up, to improve themselves or feel more in control?", "type": "A single decisive action verb phrase (one action only).", "must_not": [ "emotions", "explanations or reasons", "multiple actions", "narrative or commentary", "multi-sentence content" ] }, "this_unique_thing": { "query": "What unique benefit, feeling, or transformation do people experience after making that change?", "type": "A short noun phrase describing a concrete benefit or transformation.", "must_not": [ "full sentences", "story fragments", "emotional confession", "psychological analysis" ] } } }, { "name": "Emotional Reversal", "template": "This {unexpectedly_changed_my_feelings} about {neutral_activity}.", "description": "Shows emotional transformation and intrinsic motivation.", "variables": { "unexpectedly_changed_my_feelings": { "query": "What experiences or realizations changed how people feel, often in a surprising or positive way?", "type": "A concise emotional shift or realization (noun phrase).", "must_not": [ "multiple clauses", "behavioral descriptions", "storytelling", "explanations" ] }, "neutral_activity": { "query": "What ordinary, neutral, or routine activity were people talking about before it became meaningful?", "type": "A neutral, everyday activity (noun phrase).", "must_not": [ "emotions", "value judgments", "psychological states", "extended descriptions" ] } } }, { "name": "Collective Mastery", "template": "{ingroup} figured out how to {aspirational_outcome} in {timeframe}.", "description": "Highlights group identity, pride, and future orientation.", "variables": { "ingroup": { "query": "What group, community, or type of people are mentioned as working together or sharing values?", "type": "A group identity noun phrase.", "must_not": [ "specific individuals", "emotional language", "descriptions longer than one phrase" ] }, "aspirational_outcome": { "query": "What meaningful goal or improvement did people achieve or learn?", "type": "A short verb phrase describing a positive outcome.", "must_not": [ "reasons", "explanations", "long clauses" ] }, "timeframe": { "query": "In what period or pace did this progress occur?", "type": "A concise timeframe (number + unit).", "must_not": [ "metaphorical time", "vague or poetic phrasing", "extended descriptions" ] } } }, { "name": "Underdog Authority", "template": "The {unexpected_person} who {did_surprising_thing} is {intriguing_outcome}.", "description": "Uses status violation and admiration to evoke curiosity.", "variables": { "unexpected_person": { "query": "Who is an underestimated or ordinary person who achieved something notable?", "type": "A simple noun phrase referring to a person or role.", "must_not": [ "descriptions", "emotions", "multiple people" ] }, "did_surprising_thing": { "query": "What surprising or counterintuitive thing did this person do?", "type": "A concise verb phrase.", "must_not": [ "explanations", "multi-step actions", "emotional language" ] }, "intriguing_outcome": { "query": "What interesting result came from their action?", "type": "A short outcome phrase (adjective + noun OR verb phrase).", "must_not": [ "stories", "full sentences", "emotional framing" ] } } }, { "name": "Relational Praise Curiosity", "template": "{close_figure} said I'm finally {positive_outcome} — and here's why.", "description": "Uses relational validation and a curiosity gap to highlight personal growth.", "variables": { "close_figure": { "query": "What close social figure offers emotionally resonant validation?", "type": "A social relationship role (noun phrase).", "must_not": [ "descriptions", "emotions", "multiple figures" ] }, "positive_outcome": { "query": "What desirable transformation or identity improvement was noticed?", "type": "A short positive outcome (adjective + noun OR verb phrase).", "must_not": [ "explanations", "full sentences", "emotion words" ] } } }, { "name": "Peer Recognition Mystery", "template": "{peer_group} keep asking why I'm suddenly {new_social_trait}.", "description": "Uses implicit social proof and status elevation.", "variables": { "peer_group": { "query": "Which social circle is noticing the change?", "type": "A plural group noun phrase.", "must_not": [ "emotions", "descriptions", "specific individuals" ] }, "new_social_trait": { "query": "What desirable trait is being noticed?", "type": "A short trait phrase (adjective + noun OR behavioral noun).", "must_not": [ "explanations", "multi-clause descriptions", "emotional states" ] } } }, { "name": "Rapid Breakthrough Liberation", "template": "I {quit_negative_habit} after {short_timeframe} of using this — and I don’t even miss it.", "description": "Frames a counter-normative decision as easy and liberating.", "variables": { "quit_negative_habit": { "query": "What hard-to-quit behavior did someone give up?", "type": "A negative behavior or habit (noun phrase).", "must_not": [ "emotions", "motives", "psychological analysis", "full clauses" ] }, "short_timeframe": { "query": "What short timeframe implies rapid change?", "type": "A simple timeframe (number + unit).", "must_not": [ "vague timeframes", "emotional descriptions", "extended wording" ] } } }, { "name": "Awakened Curiosity", "template": "How my {dormant_quality} finally came back to life.", "description": "Shows revival of a valued inner quality.", "variables": { "dormant_quality": { "query": "What inner quality felt lost or suppressed before returning?", "type": "A short noun phrase naming an internal quality.", "must_not": [ "actions", "explanations", "extended descriptors" ] } } }, { "name": "Ex-Validation Twist", "template": "My ex said '{unexpected_comment}'...", "description": "Uses a surprising comment to create an identity contrast.", "variables": { "unexpected_comment": { "query": "What provocative or revealing statement did an ex make?", "type": "A short quoted sentence or phrase.", "must_not": [ "multiple sentences", "storytelling", "explanations", "long emotional descriptions" ] } } }, { "name": "Regretful Insight", "template": "Wish someone told me about {this_insight} when I was {wasting_time_behavior}.", "description": "Shows retrospective regret and transformation.", "variables": { "this_insight": { "query": "What powerful realization do people wish they discovered earlier?", "type": "A short noun phrase naming an insight, tactic, or tool.", "must_not": [ "stories", "explanations", "multiple clauses" ] }, "wasting_time_behavior": { "query": "What unproductive habit consumed time before improvement?", "type": "A negative behavior or habit (noun phrase).", "must_not": [ "emotions", "meta-analysis", "sentence-level content" ] } } }, { "name": "Empowered Intelligence", "template": "This teaches you to stop feeling bad about being {positive_trait}.", "description": "Reframes a stigmatized positive trait as empowerment.", "variables": { "positive_trait": { "query": "What undervalued trait should people feel proud of?", "type": "A positive trait or characteristic (adjective + noun OR noun phrase).", "must_not": [ "full sentences", "emotions", "extended descriptions" ] } } }, { "name": "Curiosity-Friendship Non-Tradeoff", "template": "Apparently you don't have to sacrifice {personal_value} to keep {social_group}.", "description": "Challenges tradeoff assumptions between personal growth and social belonging.", "variables": { "personal_value": { "query": "What personal value or pursuit do people fear will distance them from others?", "type": "A short noun phrase naming a value or pursuit.", "must_not": [ "sentences", "emotional language", "explanations" ] }, "social_group": { "query": "What group of friends do people want to stay connected to?", "type": "A plural group noun phrase.", "must_not": [ "emotions", "descriptions", "proper names" ] } } }, { "name": "Consequential Authority Reveal", "template": "This is what happens when {unusual_agent} finally gets {new_capability}.", "description": "Frames a surprising or status-violating outcome caused by giving agency or tools to an unexpected group.", "variables": { "unusual_agent": { "query": "Who is an unexpected or underestimated group gaining influence or tools?", "type": "A plural or collective noun phrase.", "must_not": [ "emotional labels", "descriptions", "specific individuals" ] }, "new_capability": { "query": "What new ability or leverage did they receive?", "type": "A short noun phrase.", "must_not": [ "explanations", "clauses", "emotions" ] } } }, { "name": "Anti-Conformity Shield", "template": "Found something that stops you from {forced_social_compromise}.", "description": "Positions the product as protection against social conformity pressure.", "variables": { "forced_social_compromise": { "query": "What behavior do people feel pressured into to fit in?", "type": "A negative behavioral noun phrase.", "must_not": [ "multiple behaviors", "emotional framing", "long clauses" ] } } }, { "name": "Unmasked Competence", "template": "This showed me how to stop {self_silencing_behavior}.", "description": "Describes the removal of self-suppression around intelligence or competence.", "variables": { "self_silencing_behavior": { "query": "What competence-hiding behavior were people doing before?", "type": "A short verb phrase or behavioral noun phrase.", "must_not": [ "explanations", "emotional confession", "multiple behaviors" ] } } }, { "name": "Social Permission Reframe", "template": "Apparently it's fine to {previously_discouraged_strength} sometimes.", "description": "Normalizes a trait or behavior people felt socially punished for.", "variables": { "previously_discouraged_strength": { "query": "What positive trait or behavior did people feel they had to downplay?", "type": "A short verb phrase or trait phrase.", "must_not": [ "emotional states", "multi-clause content", "explanations" ] } } }, { "name": "Intuitive Learning Compass", "template": "This teaches you to trust your gut about {learning_focus}.", "description": "Highlights regained intuition for deciding what knowledge is valuable.", "variables": { "learning_focus": { "query": "What area of learning or information choice must people rely on intuition for?", "type": "A short noun phrase.", "must_not": [ "sentences", "emotional language", "explanations" ] } } }, { "name": "Prescriptive Wisdom Signal", "template": "Honestly, this should be required for {target_group}.", "description": "Acts as a normative endorsement suggesting the product benefits an entire demographic.", "variables": { "target_group": { "query": "What demographic, age group, or category of people is being advised?", "type": "A plural or category noun phrase.", "must_not": [ "descriptions", "emotional qualifiers", "proper names" ] } } }, { "name": "Forbidden Insight Tease", "template": "{suppressing_group} don’t want you to know about {restricted_insight}.", "description": "Creates curiosity through a sense of suppressed or hidden knowledge by a group with influence or status.", "variables": { "suppressing_group": { "query": "Which group or authority is implied to be hiding information?", "type": "A plural noun phrase referring to a group.", "must_not": [ "specific individuals", "emotional framing", "long descriptions" ] }, "restricted_insight": { "query": "What powerful or useful insight is supposedly being kept from people?", "type": "A short noun phrase.", "must_not": [ "explanations", "multi-clause descriptions", "emotional labels" ] } } }, { "name": "Meta-Effectiveness Reveal", "template": "Plot twist: {unexpected_effectiveness}.", "description": "Subverts expectations by contrasting the product’s success with previous failures.", "variables": { "unexpected_effectiveness": { "query": "What concise phrase captures the surprising success compared to past disappointments?", "type": "A short noun or verb phrase expressing surprising effectiveness.", "must_not": [ "multiple sentences", "storytelling", "explicit comparisons", "emotional confession" ] } } }, { "name": "Independent Thought Activation", "template": "Found something that helps you {independent_thought_action}.", "description": "Highlights regaining the ability to think independently rather than following trends or external validation.", "variables": { "independent_thought_action": { "query": "What internal cognitive behavior did people regain or strengthen?", "type": "A short verb phrase representing autonomous thinking.", "must_not": [ "multi-step behavior", "emotional descriptors", "full clauses" ] } } } ] SCHEMAS = [ { "name": "Proven Video Script Core Block", "template": "Instead of {pain_topic_alternative}, you can use it to discover KEY IDEAS from {idea_source_list}. \n It only takes {time_value_claim}. \n It's super intuitive and you'll feel so much {intuitive_benefit}.", "description": "Preserves the exact line structure and flow of the successful script lines (00:04-00:18) while injecting fresh language via variables.", "variables": { "pain_topic_alternative": { "query": "What frustrating, time-wasting behavior are people stuck in?", "type": "A descriptive noun phrase for a negative state (e.g., 'endless brain rot', 'mindless scrolling').", "must_not": [ "full sentences", "solutions or products", "the word 'doomscrolling'" ] }, # "idea_source_list": { # "query": "What high-status sources of ideas do people want quick access to?", # "type": "A short, comma-separated list of high-status idea sources or topics.", # "must_not": [ # "full sentences", # "explanations" # ] # }, # "time_value_claim": { # "query": "What short time-for-value promise would feel impressive to them?", # "type": "A short phrase combining a specific duration with a high-value outcome.", # "must_not": [ # "vague claims", # "full sentences", # "emotional confessions" # ] # }, "intuitive_benefit": { "query": "What modern, aspirational outcome do people want more of?", "type": "A short, modern, aspirational outcome phrase.", "must_not": [ "vague feelings (happy, good)", "story fragments", "full sentences" ] } } }, ] def extract_schema_variables(comments, schema, topic, model="gpt-4.1"): system_prompt = f""" You receive: - a list of comments, - a schema with variable names and rules, - a TOPIC. TOPIC: "{topic}" Extract ALL valid values for each schema variable using ONLY words from the comments. Follow each variable’s type and must_not rules. Output JSON ONLY in this exact format: {{ "var1": [...], "var2": [...] }} Rules: - One array per variable. - No schema-name keys. - No objects, no pairs, no nesting. - No paraphrasing. - Empty array if nothing is found. """ user_payload = { "comments": comments, "schema": schema } response = client.chat.completions.create( model=model, messages=[ {"role": "system", "content": system_prompt}, {"role": "user", "content": json.dumps(user_payload, ensure_ascii=False)} ], response_format={"type": "json_object"} ) return json.loads(response.choices[0].message.content) def run_extraction_pipeline(topic): topic_suffix = topic.replace(" ", "_").lower() for schema in SCHEMAS: schema_name = schema["name"].replace(" ", "_").lower() print(f"\n🌱 Schema: {schema['name']}") for var_name in schema["variables"].keys(): var_safe = var_name.replace(" ", "_").lower() input_path = f"seeds-audible-reviews/{topic_suffix}_{schema_name}_{var_safe}.txt" if not (os.path.exists(input_path) and os.path.getsize(input_path) > 5): continue with open(input_path, "r", encoding="utf-8") as f: comments = f.read().split("\n") comments = [c.strip() for c in comments if c.strip()] print("🔍 Calling model…") result = extract_schema_variables(comments, schema, topic) values = result.get(var_name, []) output_path = f"seeds-audible-reviews/{topic_suffix}_{schema_name}_{var_safe}_keys.txt" with open(output_path, "w", encoding="utf-8") as f: if isinstance(values, list): for v in values: f.write(v.strip() + "\n") elif isinstance(values, str): f.write(values.strip() + "\n") print(f"✅ Saved {var_name} → {output_path}") if __name__ == "__main__": parser = argparse.ArgumentParser(description="Extract schema variables safely.") parser.add_argument( "--topic", type=str, required=True, help="Topic used to find comments and generate contextual extraction." ) args = parser.parse_args() run_extraction_pipeline(args.topic)