add app

app.py

@@ -0,0 +1,33 @@
+from turtle import width
+import streamlit as st
+from pathlib import Path
+from PIL import Image
+
+from variables import MAPPING_LANG_CODE_TO_TEXT, PLOT_SIZES_PER_LANG
+
+# Only need to set these here as we are add controls outside of Hydralit, to customise a run Hydralit!
+st.set_page_config(page_title="Documents sizes", layout="wide")
+
+plot_dir = Path("data/boxplot_per_ds_per_lang")
+plot_paths = list(plot_dir.iterdir())
+
+plot_paths = sorted(plot_paths)
+
+with st.sidebar:
+    st.write("Go to plot")
+    for plot_path in plot_paths:
+        plot_name = str(plot_path.name)
+        if plot_name == "colorbar.png":
+            continue
+        lang_id = plot_name.split("_")[1][:-len(".png")]
+        title = MAPPING_LANG_CODE_TO_TEXT[lang_id]
+        st.markdown(f"[{title}](#{title.replace(' ', '-').lower()})", unsafe_allow_html=True)
+
+for plot_path in plot_paths:
+    plot_name = str(plot_path.name)
+    if plot_name == "colorbar.png":
+        continue
+    lang_id = plot_name.split("_")[1][:-len(".png")]
+    st.title(MAPPING_LANG_CODE_TO_TEXT[lang_id])
+    image = Image.open(plot_path)
+    st.image(image, width=int(round(float(PLOT_SIZES_PER_LANG[lang_id]['width'])*50)))

variables.py

@@ -0,0 +1,99 @@
+MAPPING_LANG_CODE_TO_TEXT = {
+    "ar": "Arabic",
+    "ca": "Catalan",
+    "code": "code",
+    "en": "English",
+    "es": "Spanish",
+    "eu": "Basque",
+    "fr": "French",
+    "id": "Indonesian",
+    "indic-as": "Assamese",
+    "indic-bn": "Bengali",
+    "indic-gu": "Gujarati",
+    "indic-hi": "Hindi",
+    "indic-kn": "Kannada",
+    "indic-ml": "Malayalam",
+    "indic-mr": "Marathi",
+    "indic-ne": "Nepali",
+    "indic-or": "Odia",
+    "indic-pa": "Punjabi",
+    "indic-ta": "Tamil",
+    "indic-te": "Telugu",
+    "indic-ur": "Urdu",
+    "nigercongo-ak": "Akan",
+    "nigercongo-bm": "Bambara",
+    "nigercongo-fon": "Fon",
+    "nigercongo-ig": "Igbo",
+    "nigercongo-ki": "Kikuyu",
+    "nigercongo-lg": "Luganda",
+    "nigercongo-ln": "Lingala",
+    "nigercongo-nso": "Northern Sotho",
+    "nigercongo-ny": "Chi Chewa",
+    "nigercongo-rn": "Kirundi",
+    "nigercongo-rw": "Kinyarwanda",
+    "nigercongo-sn": "Chi Shona",
+    "nigercongo-st": "Sesotho",
+    "nigercongo-sw": "Swahili",
+    "nigercongo-tn": "Setswana",
+    "nigercongo-ts": "Xitsonga",
+    "nigercongo-tum": "Chi Tumbuka",
+    "nigercongo-tw": "Twi",
+    "nigercongo-wo": "Wolof",
+    "nigercongo-xh": "Xhosa",
+    "nigercongo-yo": "Yoruba",
+    "nigercongo-zu": "Isi Zulu",
+    "pt": "Portuguese",
+    "vi": "Vietnamese",
+    "zhs": "Simplified Chinese",
+    "zht": "Traditional Chinese",
+}
+
+PLOT_SIZES_PER_LANG = {
+    "indic-ta": {"width": "7.2", "num_ds": "13"},
+    "en": {"width": "19.6", "num_ds": "44"},
+    "es": {"width": "52.0", "num_ds": "125"},
+    "indic-kn": {"width": "5.2", "num_ds": "8"},
+    "zht": {"width": "2.8", "num_ds": "2"},
+    "nigercongo-ki": {"width": "2.4", "num_ds": "1"},
+    "indic-pa": {"width": "5.6", "num_ds": "9"},
+    "vi": {"width": "10.0", "num_ds": "20"},
+    "zhs": {"width": "8.8", "num_ds": "17"},
+    "fr": {"width": "10.8", "num_ds": "22"},
+    "eu": {"width": "7.6000000000000005", "num_ds": "14"},
+    "indic-te": {"width": "6.800000000000001", "num_ds": "12"},
+    "indic-hi": {"width": "10.0", "num_ds": "20"},
+    "pt": {"width": "9.600000000000001", "num_ds": "19"},
+    "indic-bn": {"width": "8.4", "num_ds": "16"},
+    "indic-mr": {"width": "6.4", "num_ds": "11"},
+    "indic-gu": {"width": "6.0", "num_ds": "10"},
+    "ca": {"width": "10.0", "num_ds": "20"},
+    "id": {"width": "12.4", "num_ds": "26"},
+    "ar": {"width": "12.0", "num_ds": "25"},
+    "indic-or": {"width": "5.6", "num_ds": "9"},
+    "indic-ur": {"width": "7.2", "num_ds": "13"},
+    "nigercongo-ig": {"width": "2.4", "num_ds": "1"},
+    "indic-as": {"width": "4.4", "num_ds": "6"},
+    "indic-ml": {"width": "6.800000000000001", "num_ds": "12"},
+    "nigercongo-ny": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-tw": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-rn": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-st": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-yo": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-ak": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-lg": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-bm": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-wo": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-ln": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-nso": {"width": "2.4", "num_ds": "1"},
+    "code": {"width": "2.8", "num_ds": "2"},
+    "indic-ne": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-ts": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-zu": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-sn": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-sw": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-tum": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-tn": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-xh": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-rw": {"width": "2.4", "num_ds": "1"},
+    "nigercongo-fon": {"width": "2.4", "num_ds": "1"},
+}