From 07be9ea6e34e88b98906d4ee8fd9f5a858285331 Mon Sep 17 00:00:00 2001
From: Jake Poznanski <jakep@allenai.org>
Date: Mon, 10 Mar 2025 21:55:33 +0000
Subject: [PATCH] More math testing

---
 olmocr/bench/benchmark.py              |  4 ++++
 olmocr/bench/katex/render.py           | 12 ++++++++++++
 olmocr/bench/sample_data/dataset.jsonl |  5 +++++
 3 files changed, 21 insertions(+)

diff --git a/olmocr/bench/benchmark.py b/olmocr/bench/benchmark.py
index f21b39a..a7da1f1 100644
--- a/olmocr/bench/benchmark.py
+++ b/olmocr/bench/benchmark.py
@@ -21,6 +21,7 @@ import sys
 from typing import Dict, List, Tuple, Optional
 
 from .tests import BasePDFTest, BaselineTest, load_tests
+from .katex.render import clear_cache_dir
 from .utils import calculate_bootstrap_ci, perform_permutation_test
 
 def evaluate_candidate(
@@ -152,6 +153,9 @@ def main():
     ci_level = args.confidence_level
     pdf_folder = os.path.join(input_folder, "pdfs")
 
+    # Clear equation cache directory
+    clear_cache_dir()
+
     # Check that the pdfs folder exists
     if not os.path.exists(pdf_folder):
         print("Error: /pdfs folder must exist in your data directory.", file=sys.stderr)
diff --git a/olmocr/bench/katex/render.py b/olmocr/bench/katex/render.py
index 7fa41fd..4708a6d 100644
--- a/olmocr/bench/katex/render.py
+++ b/olmocr/bench/katex/render.py
@@ -16,6 +16,7 @@ import hashlib
 import pathlib
 import json
 import re
+import shutil
 from dataclasses import dataclass
 from typing import List
 import unittest
@@ -55,6 +56,17 @@ def get_cache_dir():
     cache_dir.mkdir(parents=True, exist_ok=True)
     return cache_dir
 
+
+def clear_cache_dir():
+    """
+    Clear all files and subdirectories in the cache directory.
+    """
+    cache_dir = get_cache_dir()
+    if cache_dir.exists() and cache_dir.is_dir():
+        shutil.rmtree(cache_dir)
+        cache_dir.mkdir(parents=True, exist_ok=True)  # Recreate the empty directory
+
+
 def render_equation(
     equation, 
     bg_color="white",
diff --git a/olmocr/bench/sample_data/dataset.jsonl b/olmocr/bench/sample_data/dataset.jsonl
index 820cbf5..de12ff2 100644
--- a/olmocr/bench/sample_data/dataset.jsonl
+++ b/olmocr/bench/sample_data/dataset.jsonl
@@ -67,3 +67,8 @@
 {"pdf": "mathfuncs_colswitch.pdf", "page": 1, "id": "mathfuncscol_00", "type": "order", "before": "Euler's Identity", "after": "Pythagorean Theorem"}
 {"pdf": "mathfuncs_colswitch.pdf", "page": 1, "id": "mathfuncscol_01", "type": "order", "before": "Pythagorean Theorem", "after": "The Fundamental Theorem of Calculus"}
 {"pdf": "mathfuncs_colswitch.pdf", "page": 1, "id": "mathfuncscol_02", "type": "order", "before": "The Fundamental Theorem of Calculus", "after": "Maxwell's Equations"}
+
+{"pdf": "math_2503_04086.pdf", "page": 1, "id": "math_2503_04086_00", "type": "math", "math": "\\lambda_{g}=\\sum_{s \\in S} \\zeta_{n}^{\\psi(g s)}=\\sum_{i=1}^{k}\\left[\\sum_{s, R s=\\mathcal{I}_{i}} \\zeta_{n}^{\\varphi(g s)}\\right]"}
+{"pdf": "math_2503_04086.pdf", "page": 1, "id": "math_2503_04086_01", "type": "math", "math": "\\lambda_{g}=\\lambda_{g^{\\prime}}"}
+{"pdf": "math_2503_04086.pdf", "page": 1, "id": "math_2503_04086_02", "type": "math", "math": "u \\in\\left(R / \\operatorname{Ann}_{R}\\left(x_{i}\\right)\\right)^{\\times}"}
+{"pdf": "math_2503_04086.pdf", "page": 1, "id": "math_2503_04086_03", "type": "math", "math": "\\lambda_{g}=\\sum_{i=1}^{k} c\\left(g, R / \\operatorname{Ann}_{R}\\left(x_{i}\\right)\\right)"}