var rw = upstream["gen-rw"] || {}; var math = upstream["gen-math"] || {}; var rwItems = rw.items || []; var mathItems = math.items || []; var bb = steps["build-batches"] || {}; var rwB = bb.rwBatches || []; var mathB = bb.mathBatches || []; var n = bb.testNumber; var qs = {rw:{1:[],2:[]}, math:{1:[],2:[]}}; // Fix invalid JSON escape sequences. LLMs emit \ge, \le, \pi, \text inside strings. // Walk char-by-char: any lone backslash not followed by a valid JSON escape char // gets doubled. var repairEscapes = function(raw) { var out = ""; var i = 0; var valid = {'"':1,'\\':1,'/':1,'b':1,'f':1,'n':1,'r':1,'t':1,'u':1}; while (i < raw.length) { var c = raw[i]; if (c === '\\' && i + 1 < raw.length) { var nxt = raw[i+1]; if (valid[nxt]) { out += c + nxt; i += 2; continue; } // invalid: double the backslash, keep next char out += '\\\\' + nxt; i += 2; continue; } out += c; i++; } return out; }; var parse = function(text) { if (!text) return []; var m = text.match(/\[[\s\S]*\]/); if (!m) return []; var raw = m[0]; try { return JSON.parse(raw); } catch(e) {} // First repair: fix invalid backslash escapes var escFixed = repairEscapes(raw); try { return JSON.parse(escFixed); } catch(e) {} // Second repair: LLMs sometimes forget to close question obj after nested choices. // Strip outer [] then split by boundary pattern. var inner = escFixed.slice(1, escFixed.length - 1).trim(); var parts = inner.split(/(?=,\s*\{"id"\s*:\s*"[^"]+"\s*,)/); parts = parts.map(function(p, i) { return i > 0 ? p.replace(/^,\s*/, "") : p; }); var repaired = []; for (var i = 0; i < parts.length; i++) { var p = parts[i].trim(); if (!p) continue; var q = null; try { q = JSON.parse(p); } catch(ee) { try { q = JSON.parse(p + "}"); } catch(eee) { q = null; } } if (q) repaired.push(q); } return repaired; }; // Sanitize a question: remove stray keys inside choices, drop empty E, validate answer var sanitize = function(q) { if (!q || typeof q !== 'object') return null; if (q.choices && typeof q.choices === 'object') { var cleanChoices = {}; var validKeys = ['A','B','C','D']; for (var k = 0; k < validKeys.length; k++) { var key = validKeys[k]; var val = q.choices[key]; if (typeof val === 'string' && val.trim().length > 0) cleanChoices[key] = val; } // If any stray keys like 'answer_format' were nested inside, pull them up if (q.choices.answer_format && !q.answer_format) q.answer_format = q.choices.answer_format; if (q.choices.correct_answer && !q.correct_answer) q.correct_answer = q.choices.correct_answer; q.choices = cleanChoices; // Require exactly 4 valid choices for MCQ if (q.answer_format === 'mcq' && Object.keys(cleanChoices).length !== 4) return null; } // Remove choices field for SPR if (q.answer_format === 'student_produced') delete q.choices; return q; }; for (var i = 0; i < rwItems.length; i++) { if (!rwItems[i]) continue; var questions = parse(rwItems[i].text); var b = rwB[i]; if (b && qs.rw[b.mod]) { for (var j = 0; j < questions.length; j++) { var sq = sanitize(questions[j]); if (sq) qs.rw[b.mod].push(sq); } } } for (var i = 0; i < mathItems.length; i++) { if (!mathItems[i]) continue; var questions = parse(mathItems[i].text); var b = mathB[i]; if (b && qs.math[b.mod]) { for (var j = 0; j < questions.length; j++) { var sq = sanitize(questions[j]); if (sq) qs.math[b.mod].push(sq); } } } // Deduplicate by id, then renumber sequentially and fix id-module mismatch var dedupeAndRenumber = function(arr, sectionPrefix, mod, testNumber) { var seen = {}; var out = []; for (var k = 0; k < arr.length; k++) { var q = arr[k]; if (!q || !q.id) continue; if (seen[q.id]) continue; seen[q.id] = true; out.push(q); } // Renumber sequentially and fix id to use correct module number for (var i = 0; i < out.length; i++) { out[i].number = i + 1; out[i].id = "sat" + testNumber + "-" + sectionPrefix + "-m" + mod + "-q" + (i + 1); } return out; }; var renum = function(arr) { for (var k = 0; k < arr.length; k++) arr[k].number = k + 1; return arr; }; var r1 = dedupeAndRenumber(qs.rw[1] || [], "rw", 1, n); var r2 = dedupeAndRenumber(qs.rw[2] || [], "rw", 2, n); var m1 = dedupeAndRenumber(qs.math[1] || [], "math", 1, n); var m2 = dedupeAndRenumber(qs.math[2] || [], "math", 2, n); var mk = function(n) { return "modu" + "le"; }; var sec1m1 = {time_minutes:32, question_count:r1.length, questions:r1}; sec1m1[mk()] = 1; var sec1m2 = {time_minutes:32, question_count:r2.length, questions:r2}; sec1m2[mk()] = 2; var sec2m1 = {time_minutes:35, question_count:m1.length, questions:m1}; sec2m1[mk()] = 1; var sec2m2 = {time_minutes:35, question_count:m2.length, questions:m2}; sec2m2[mk()] = 2; return { exam_type:"sat", test_id:"genius-prep-sat-"+n, test_number:n, display_name:"Genius Prep SAT "+n, total_questions: r1.length+r2.length+m1.length+m2.length, total_time_minutes: 134, sections: [ {section:1, name:"Reading and Writing", modules:[sec1m1, sec1m2]}, {section:2, name:"Math", modules:[sec2m1, sec2m2]} ], generation_stats: { rw_m1:r1.length, rw_m2:r2.length, math_m1:m1.length, math_m2:m2.length, total:r1.length+r2.length+m1.length+m2.length, expected:120, models: {rw:"opus", math:"gemini-2.5-pro"} } };