Add tf-idf notebook

move data files into their own directory

Author: adetorcy

data/edrm_subset.sqlite3.gz

@@ -70,8 +70,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "src = Path(\"RevisedEDRMv1_Complete/albert_meyers/\") # for PST files\n",
-    "# src = Path(\"httpd-users\")  # for mbox files"
+    "src = Path(\"data/RevisedEDRMv1_Complete/albert_meyers/\") # for PST files\n",
+    "# src = Path(\"data/httpd-users\")  # for mbox files"
    ]
   },
   {
@@ -319,7 +319,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.5"
+   "version": "3.8.2"
   },
   "pycharm": {
    "stem_cell": {

entities.ipynb

@@ -61,7 +61,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "location = Path('RevisedEDRMv1_Complete/andrea_ring/')"
+    "location = Path('data/RevisedEDRMv1_Complete/andrea_ring/')"
    ]
   },
   {
@@ -113,7 +113,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.5"
+   "version": "3.8.2"
   }
  },
  "nbformat": 4,

export_eml.ipynb

@@ -55,7 +55,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "PST_FILE = Path(\"RevisedEDRMv1_Complete/andrea_ring/andrea_ring_000_1_1.pst\")"
+    "PST_FILE = Path(\"data/RevisedEDRMv1_Complete/andrea_ring/andrea_ring_000_1_1.pst\")"
    ]
   },
   {
@@ -1373,7 +1373,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.5"
+   "version": "3.8.2"
   },
   "pycharm": {
    "stem_cell": {

message_viewer.ipynb

@@ -1,9 +1,9 @@
 #!/bin/bash
 
-PST_DATA_DIR="RevisedEDRMv1_Complete"
+PST_DATA_DIR="data/RevisedEDRMv1_Complete"
 ZIPPED_PST_FILES=(albert_meyers.zip andrea_ring.zip)
 
-MBOX_DATA_DIR="httpd-users"
+MBOX_DATA_DIR="data/httpd-users"
 
 mkdir -p $PST_DATA_DIR $MBOX_DATA_DIR
 

postBuild

@@ -53,7 +53,7 @@
    "outputs": [],
    "source": [
     "# Edit as appropriate\n",
-    "CACHED_ENRON_DATA_DIR = Path(\"RevisedEDRMv1_Complete\")\n",
+    "CACHED_ENRON_DATA_DIR = Path(\"data/RevisedEDRMv1_Complete\")\n",
     "\n",
     "# Generate the list of files to know how many there are\n",
     "files = sorted(CACHED_ENRON_DATA_DIR.glob('**/*.pst'))"
@@ -184,7 +184,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.5"
+   "version": "3.8.2"
   }
  },
  "nbformat": 4,

pst_walk.ipynb

@@ -7,3 +7,4 @@ jupyterlab
 ipywidgets
 matplotlib
 sklearn
+terminado==0.8.3        # for mybinder.org

requirements.in

@@ -14,7 +14,7 @@ blessed==1.17.8           # via enlighten, libratom
 blis==0.4.1               # via libratom, spacy, thinc
 catalogue==1.0.0          # via libratom, spacy, thinc
 certifi==2020.6.20        # via libratom, matplotlib, requests
-cffi==1.14.2              # via argon2-cffi
+cffi==1.14.3              # via argon2-cffi
 chardet==3.0.4            # via libratom, requests
 click-log==0.3.2          # via libratom
 click==7.1.2              # via click-log, libratom
@@ -39,7 +39,7 @@ json5==0.9.5              # via jupyterlab-server
 jsonschema==3.2.0         # via jupyterlab-server, libratom, nbformat
 jupyter-client==6.1.7     # via ipykernel, nbclient, notebook
 jupyter-core==4.6.3       # via jupyter-client, nbconvert, nbformat, notebook
-jupyterlab-pygments==0.1.1  # via nbconvert
+jupyterlab-pygments==0.1.2  # via nbconvert
 jupyterlab-server==1.2.0  # via jupyterlab
 jupyterlab==2.2.8         # via -r requirements.in
 kiwisolver==1.2.0         # via matplotlib
@@ -50,9 +50,9 @@ matplotlib==3.3.2         # via -r requirements.in
 mistune==0.8.4            # via nbconvert
 murmurhash==1.0.2         # via libratom, preshed, spacy, thinc
 nbclient==0.5.0           # via nbconvert
-nbconvert==6.0.4          # via notebook
+nbconvert==6.0.6          # via notebook
 nbformat==5.0.7           # via ipywidgets, nbclient, nbconvert, notebook
-nest-asyncio==1.4.0       # via nbclient
+nest-asyncio==1.4.1       # via nbclient
 notebook==6.1.4           # via jupyterlab, jupyterlab-server, widgetsnbextension
 numpy==1.19.0             # via blis, libratom, matplotlib, pandas, scikit-learn, scipy, spacy, thinc
 packaging==20.4           # via bleach, libratom
@@ -88,7 +88,7 @@ sqlalchemy==1.3.18        # via -r requirements.in, libratom
 srsly==1.0.2              # via libratom, spacy, thinc
 striprtf==0.0.10          # via libratom
 tabulate==0.8.7           # via libratom
-terminado==0.9.1          # via notebook
+terminado==0.8.3          # via -r requirements.in, notebook
 testpath==0.4.4           # via nbconvert
 thinc==7.4.1              # via libratom, spacy
 threadpoolctl==2.1.0      # via scikit-learn