Setup
Build the Table
- Write to File
Diagnostics
- Missing or Extra XPATHS
- Failes URLs
XML Table Structure
- Create Network Edgelists
- Visualize Tree Structure
XPATH Summary

Setup

Load Packages

library( irs990efile )  # functions for parsing efile XMLs
library( dplyr )        # data wrangling
library( purrr )        # data wrangling 
library( pander )       # formatting
library( knitr )        # formatting
library( DT )           # table printing
library( httr )         # web requests 
library( xmltools )     # xml utilities
library( xml2 )         # xml utilities
library( XML )          # xml utilities 
library( data.tree )    # network visualization
library( networkD3 )    # network visualization
library( igraph )       # network visualization

source( "R/rdb-functions-v2.R")
source( "R/utils.R")

Data

head( index )

The Concordance File

The concordance is an RDB to XML Crosswalk that maps all xpaths onto a specified set of variables. It is located in the irs990efile package.

The full data dictionary is available on GitHub.

DATA DICTIONARY

concordance %>% 
  filter( rdb_table == table.name )

xpath	id
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/DescriptionEntPrimaryActivity	1
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/NameOfEntity/BusinessNameLine1	2
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/NameOfEntity/BusinessNameLine2	3
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/OfficersEtcProfitOrOwnership	4
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/OrganizationProfitOrOwnership	5
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/PhysicansProfitOrOwnership	6
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine1	7
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine1Txt	8
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine2	9
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine2Txt	10
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/OfcrEtcProfitOrOwnershipPct	11
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/OrgProfitOrOwnershipPct	12
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/PhysiciansProfitOrOwnershipPct	13
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/PrimaryActivitiesTxt	14

xpaths <- as.character( t.xpaths$xpath )
el <- create_edgelist_v1( xpaths )
nd <- FromDataFrameNetwork( network=el )
print( nd )

##                                         levelName
## 1  Return                                        
## 2   °--ReturnData                                
## 3       °--IRS990ScheduleH                       
## 4           ¦--Form990ScheduleHPartIV            
## 5           ¦   ¦--DescriptionEntPrimaryActivity 
## 6           ¦   ¦--NameOfEntity                  
## 7           ¦   ¦   ¦--BusinessNameLine1         
## 8           ¦   ¦   °--BusinessNameLine2         
## 9           ¦   ¦--OfficersEtcProfitOrOwnership  
## 10          ¦   ¦--OrganizationProfitOrOwnership 
## 11          ¦   °--PhysicansProfitOrOwnership    
## 12          °--ManagementCoAndJntVenturesGrp     
## 13              ¦--EntityName                    
## 14              ¦   ¦--BusinessNameLine1         
## 15              ¦   ¦--BusinessNameLine1Txt      
## 16              ¦   ¦--BusinessNameLine2         
## 17              ¦   °--BusinessNameLine2Txt      
## 18              ¦--OfcrEtcProfitOrOwnershipPct   
## 19              ¦--OrgProfitOrOwnershipPct       
## 20              ¦--PhysiciansProfitOrOwnershipPct
## 21              °--PrimaryActivitiesTxt

SetGraphStyle( nd, rankdir = "LR")
SetEdgeStyle( nd, arrowhead = "vee", color = "grey20", penwidth = 2 )
SetNodeStyle( nd, 
              style = "filled,rounded", 
              shape = "box", 
              fillcolor = "LightBlue", 
              fontname = "helvetica", 
              fontcolor="black",
              tooltip = GetDefaultTooltip )

# SetNodeStyle(acme$IT, fillcolor = "LightBlue", penwidth = "5px")
plot( nd )

# old version
find_group_names( table.name )

## [1] "//ManagementCoAndJntVenturesGrp" "//Form990ScheduleHPartIV"

# new version
find_table_headers( table.name )

##  [1] "//ManagementCoAndJntVenturesGrp/EntityName"                    
##  [2] "//ManagementCoAndJntVenturesGrp/PrimaryActivitiesTxt"          
##  [3] "//ManagementCoAndJntVenturesGrp/OrgProfitOrOwnershipPct"       
##  [4] "//ManagementCoAndJntVenturesGrp/OfcrEtcProfitOrOwnershipPct"   
##  [5] "//ManagementCoAndJntVenturesGrp/PhysiciansProfitOrOwnershipPct"
##  [6] "//Form990ScheduleHPartIV/NameOfEntity"                         
##  [7] "//Form990ScheduleHPartIV/DescriptionEntPrimaryActivity"        
##  [8] "//Form990ScheduleHPartIV/OrganizationProfitOrOwnership"        
##  [9] "//Form990ScheduleHPartIV/OfficersEtcProfitOrOwnership"         
## [10] "//Form990ScheduleHPartIV/PhysicansProfitOrOwnership"

# manual
if( ! is.null(table.headers) )
{table.headers}

Build the Table

# create new folder
dir.create( folder.name )
setwd( folder.name )

# create sample
set.seed( 1234 )
sample.index <- sample_n( index, sample.size )
write.csv( index, "sample-index.csv", row.names=F )
sample.urls <- sample.index$URL

# erase existing log files
file.create("XPATH-LOG.txt") 
file.create("FAIL-LOG.txt")


start.build.time <- Sys.time()    # --------------------

results.list <- list()
xpath.list <- list()


for( i in 1:length( sample.urls ) )
{
  
  url <- sample.urls[i]
  
  results.list[[i]] <- 
    build_rdb_table_v2( url, 
                        table.name, 
                        table.headers=table.headers )
  
  xpath.list[[i]]   <- 
    get_table_xpaths( url, 
                      table.name, 
                      table.headers=table.headers )
  
  if( i %% 100 == 0 ){ print(i) }
  
}


end.build.time <- Sys.time()      # --------------------


df <- dplyr::bind_rows( results.list )


fail.log <- readLines( "FAIL-LOG.txt" )
all.xpaths <- xpath.list %>% unlist() %>% unique() %>% sort()
cat( all.xpaths, sep="\n", file="XPATH-LOG.txt" )

TABLE: SH-P04-T01-COMPANY-JOINT-VENTURES

## [1] "NUMBER OF RETURNS PROCESSED:   1000"
## [1] "TOTAL NUMBER OF TABLE ROWS:   535"

TOTAL RUN TIME:

## Time difference of 21.51521 mins
## Time difference of -0.3585868 hours

Write to File

write.csv( df, paste0( folder.name, "/", table.name, ".csv" ), row.names=F )
head(df)

Are all names standardized?

names(df)

##  [1] "OBJECT_ID"                     "EIN"                          
##  [3] "NAME"                          "TAXYR"                        
##  [5] "FORMTYPE"                      "URL"                          
##  [7] "SH_04_CO_ENTITY_ACT_DESC"      "SH_04_CO_PROFIT_PCT_ORG"      
##  [9] "SH_04_CO_PROFIT_PCT_PHYSICIAN" "SH_04_CO_ENTITY_NAME_L1"      
## [11] "SH_04_CO_PROFIT_PCT_OFF"       "SH_04_CO_ENTITY_NAME_L1.1"

Diagnostics

Missing or Extra XPATHS

Check for xpaths that were used in the tables but do not exist in the Concordance file.

c.table <- dplyr::filter( concordance, rdb_table == table.name ) 
xpath.concordance <- c.table$xpath
setdiff( all.xpaths, xpath.concordance )

## [1] ""                                                                           
## [2] "/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName"

Capture to update the Concordance file.

these <- setdiff( all.xpaths, xpath.concordance )
cat( these, sep="\n", file=paste0( folder.name, "/MISSING-XPATHS.txt" ) )

Failes URLs

Check the status of failed URLs.

if( length(fail.log) > 0 )
{ 
  sapply( fail.log, get_message ) %>% 
  table() %>% 
  pander() 
} else {
  print("ALL URLs ARE VALID")
}

## [1] "ALL URLs ARE VALID"

XML Table Structure

Create Network Edgelists

The create_edgelist() functions v1 and v2 are in R/utils.R.

V1 retains the root nodes Return and ReturnData.

V2 drops the root nodes and uses form versions “990” and “990EZ” as the root to visualize the sets of tables separately.

# KEEP ROOT NOTES Return & ReturnData
el <- create_edgelist_v1( all.xpaths )

Edgelist format:

V1	V2
Return	ReturnData
ReturnData	IRS990
IRS990	Form990PartVIISectionAGrp
Form990PartVIISectionAGrp	PersonNm
Form990PartVIISectionAGrp	TitleTxt
Form990PartVIISectionAGrp	AverageHoursPerWeekRt
Form990PartVIISectionAGrp	IndividualTrusteeOrDirectorInd
Form990PartVIISectionAGrp	OfficerInd
Form990PartVIISectionAGrp	ReportableCompFromOrgAmt
Form990PartVIISectionAGrp	ReportableCompFromRltdOrgAmt

Visualize Tree Structure

# library( data.tree )
nd <- FromDataFrameNetwork( network=el )
print( nd )

##                                         levelName
## 1  Return                                        
## 2   °--ReturnData                                
## 3       °--IRS990ScheduleH                       
## 4           °--ManagementCoAndJntVenturesGrp     
## 5               ¦--EntityName                    
## 6               ¦   °--BusinessNameLine1Txt      
## 7               ¦--OfcrEtcProfitOrOwnershipPct   
## 8               ¦--OrgProfitOrOwnershipPct       
## 9               ¦--PhysiciansProfitOrOwnershipPct
## 10              °--PrimaryActivitiesTxt

SetGraphStyle( nd, rankdir = "LR")
SetEdgeStyle( nd, arrowhead = "vee", color = "grey20", penwidth = 2 )
SetNodeStyle( nd, 
              style = "filled,rounded", 
              shape = "box", 
              fillcolor = "LightBlue", 
              fontname = "helvetica", 
              fontcolor="black",
              tooltip = GetDefaultTooltip )

# SetNodeStyle(acme$IT, fillcolor = "LightBlue", penwidth = "5px")
plot( nd )

# library( networkD3 )
# net <- ToDataFrameNetwork( nd, "name" )
# simpleNetwork( net, fontSize = 12 )
simpleNetwork( el, fontSize = 12 )

#plot with networkD3
radnet <- ToListExplicit( nd, unname = TRUE )
radialNetwork( radnet, fontSize = 12 )

XPATH Summary

Unique Xpaths for this RDB Table

t.xpaths <- 
  concordance %>%
  filter( rdb_table == table.name ) %>% 
  select( xpath ) %>% 
  unique() %>% 
  arrange( xpath ) %>% 
  mutate( id=row_number() ) 

t.xpaths %>% 
  kable( align="l" )

xpath	id
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/DescriptionEntPrimaryActivity	1
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/NameOfEntity/BusinessNameLine1	2
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/NameOfEntity/BusinessNameLine2	3
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/OfficersEtcProfitOrOwnership	4
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/OrganizationProfitOrOwnership	5
/Return/ReturnData/IRS990ScheduleH/Form990ScheduleHPartIV/PhysicansProfitOrOwnership	6
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine1	7
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine1Txt	8
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine2	9
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/EntityName/BusinessNameLine2Txt	10
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/OfcrEtcProfitOrOwnershipPct	11
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/OrgProfitOrOwnershipPct	12
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/PhysiciansProfitOrOwnershipPct	13
/Return/ReturnData/IRS990ScheduleH/ManagementCoAndJntVenturesGrp/PrimaryActivitiesTxt	14

Grouping Variables

Identify all unique groups used in xpaths.

if( is.null(table.headers) )
{ find_table_headers( table.name ) } else
table.headers

##  [1] "//ManagementCoAndJntVenturesGrp/EntityName"                    
##  [2] "//ManagementCoAndJntVenturesGrp/PrimaryActivitiesTxt"          
##  [3] "//ManagementCoAndJntVenturesGrp/OrgProfitOrOwnershipPct"       
##  [4] "//ManagementCoAndJntVenturesGrp/OfcrEtcProfitOrOwnershipPct"   
##  [5] "//ManagementCoAndJntVenturesGrp/PhysiciansProfitOrOwnershipPct"
##  [6] "//Form990ScheduleHPartIV/NameOfEntity"                         
##  [7] "//Form990ScheduleHPartIV/DescriptionEntPrimaryActivity"        
##  [8] "//Form990ScheduleHPartIV/OrganizationProfitOrOwnership"        
##  [9] "//Form990ScheduleHPartIV/OfficersEtcProfitOrOwnership"         
## [10] "//Form990ScheduleHPartIV/PhysicansProfitOrOwnership"

Groups can then be used to extract the 1:M RDB tables:

if( is.null(table.headers) )
{ table.headers <- find_table_headers( table.name ) }
demo.url <- "https://nccs-efile.s3.us-east-1.amazonaws.com/xml/201913079349300206_public.xml"
doc <- xml2::read_xml( demo.url )
xml2::xml_ns_strip( doc )
df.demo <- get_table_v2( doc, table.name, table.headers )
df.demo

## NULL

Standardize Variable Names

Mapping of xml variable names to concordance variable names:

get_var_map( table.name=table.name )

A table with proper names:

v.map <- get_var_map( table.name=table.name )

if( ! is.null(df.demo) )
{re_name( df.demo, v.map ) %>% as.data.frame()}

Print Table Structure

Demo of functions - demo data only.

[[1]]
<Form990PartVIISectionAGrp>
  <PersonNm>
    {text}
  <TitleTxt>
    {text}
  <AverageHoursPerWeekRt>
    {text}
  <IndividualTrusteeOrDirectorInd>
    {text}
  <OfficerInd>
    {text}
  <ReportableCompFromOrgAmt>
    {text}
  <ReportableCompFromRltdOrgAmt>
    {text}
  <OtherCompensationAmt>
    {text}

node.list[[1]] %>% xml_parent() %>% xmltools::xml_view_trees()

 (1) ------------

 +-- Form990PartVIISectionAGrp
  +-- PersonNm
  +-- TitleTxt
  +-- AverageHoursPerWeekRt
  +-- IndividualTrusteeOrDirectorInd
  +-- OfficerInd
  +-- ReportableCompFromOrgAmt
  +-- ReportableCompFromRltdOrgAmt
  ++-- OtherCompensationAmt


 (2) ------------

 +-- Form990PartVIISectionAGrp
  +-- PersonNm
  +-- TitleTxt
  +-- AverageHoursPerWeekRt
  +-- IndividualTrusteeOrDirectorInd
  +-- OfficerInd
  +-- ReportableCompFromOrgAmt
  +-- ReportableCompFromRltdOrgAmt
  ++-- OtherCompensationAmt

TABLE NAME: SH-P04-T01-COMPANY-JOINT-VENTURES

Last compiled on 04 April, 2023