About Me Blog
A tutorial on tidy cross-validation with R Analyzing NetHack data, part 1: What kills the players Analyzing NetHack data, part 2: What players kill the most Building a shiny app to explore historical newspapers: a step-by-step guide Classification of historical newspapers content: a tutorial combining R, bash and Vowpal Wabbit, part 1 Classification of historical newspapers content: a tutorial combining R, bash and Vowpal Wabbit, part 2 Curly-Curly, the successor of Bang-Bang Dealing with heteroskedasticity; regression with robust standard errors using R Easy time-series prediction with R: a tutorial with air traffic data from Lux Airport Exporting editable plots from R to Powerpoint: making ggplot2 purrr with officer Fast food, causality and R packages, part 1 Fast food, causality and R packages, part 2 For posterity: install {xml2} on GNU/Linux distros Forecasting my weight with R From webscraping data to releasing it as an R package to share with the world: a full tutorial with data from NetHack Get text from pdfs or images using OCR: a tutorial with {tesseract} and {magick} Getting data from pdfs using the pdftools package Getting the data from the Luxembourguish elections out of Excel Going from a human readable Excel file to a machine-readable csv with {tidyxl} Historical newspaper scraping with {tesseract} and R How Luxembourguish residents spend their time: a small {flexdashboard} demo using the Time use survey data Imputing missing values in parallel using {furrr} Intermittent demand, Croston and Die Hard Looking into 19th century ads from a Luxembourguish newspaper with R Making sense of the METS and ALTO XML standards Manipulate dates easily with {lubridate} Manipulating strings with the {stringr} package Maps with pie charts on top of each administrative division: an example with Luxembourg's elections data Missing data imputation and instrumental variables regression: the tidy approach Modern R with the tidyverse is available on Leanpub Objects types and some useful R functions for beginners Pivoting data frames just got easier thanks to `pivot_wide()` and `pivot_long()` R or Python? Why not both? Using Anaconda Python within R with {reticulate} Searching for the optimal hyper-parameters of an ARIMA model in parallel: the tidy gridsearch approach Some fun with {gganimate} Split-apply-combine for Maximum Likelihood Estimation of a linear model Statistical matching, or when one single data source is not enough The best way to visit Luxembourguish castles is doing data science + combinatorial optimization The never-ending editor war (?) The year of the GNU+Linux desktop is upon us: using user ratings of Steam Play compatibility to play around with regex and the tidyverse Using Data Science to read 10 years of Luxembourguish newspapers from the 19th century Using a genetic algorithm for the hyperparameter optimization of a SARIMA model Using cosine similarity to find matching documents: a tutorial using Seneca's letters to his friend Lucilius Using linear models with binary dependent variables, a simulation study Using the tidyverse for more than data manipulation: estimating pi with Monte Carlo methods What hyper-parameters are, and what to do with them; an illustration with ridge regression {disk.frame} is epic {pmice}, an experimental package for missing data imputation in parallel using {mice} and {furrr} Building formulae Functional peace of mind Get basic summary statistics for all the variables in a data frame Getting {sparklyr}, {h2o}, {rsparkling} to work together and some fun with bash Importing 30GB of data into R with sparklyr Introducing brotools It's lists all the way down It's lists all the way down, part 2: We need to go deeper Keep trying that api call with purrr::possibly() Lesser known dplyr 0.7* tricks Lesser known dplyr tricks Lesser known purrr tricks Make ggplot2 purrr Mapping a list of functions to a list of datasets with a list of columns as arguments Predicting job search by training a random forest on an unbalanced dataset Teaching the tidyverse to beginners Why I find tidyeval useful tidyr::spread() and dplyr::rename_at() in action Easy peasy STATA-like marginal effects with R Functional programming and unit testing for data munging with R available on Leanpub How to use jailbreakr My free book has a cover! Work on lists of datasets instead of individual datasets by using functional programming Method of Simulated Moments with R New website! Nonlinear Gmm with R - Example with a logistic regression Simulated Maximum Likelihood with R Bootstrapping standard errors for difference-in-differences estimation with R Careful with tryCatch Data frame columns as arguments to dplyr functions Export R output to a file I've started writing a 'book': Functional programming and unit testing for data munging with R Introduction to programming econometrics with R Merge a list of datasets together Object Oriented Programming with R: An example with a Cournot duopoly R, R with Atlas, R with OpenBLAS and Revolution R Open: which is fastest? Read a lot of datasets at once with R Unit testing with R Update to Introduction to programming econometrics with R Using R as a Computer Algebra System with Ryacas

Making sense of the METS and ALTO XML standards

Last week I wrote a blog post where I analyzed one year of newspapers ads from 19th century newspapers. The data is made available by the national library of Luxembourg. In this blog post, which is part 1 of a 2 part series, I extract data from the 257gb archive, which contains 10 years of publications of the L’Union, another 19th century Luxembourguish newspaper written in French. As I explained in the previous post, to make life easier to data scientists, the national library also included ALTO and METS files (which are a XML files used to describe the layout and contents of physical text sources, such as pages of a book or newspaper) which can be easily parsed by R.

This is how a ALTO file looks like:

Each page of the newspaper of a given day has one ALTO file. This is how a METS file looks like:

For each daily issue of the newspaper, there is a METS file. So 1 METS file for 4 ALTO files.

In my last blog post, I only extracted the words from the ALTO file (red rectangles of the first screenshot) and did not touch the METS file. The problem of doing this is that I get all the words for each page, without knowing which come from the same article. If I want to know which words come from the same article, I need to use the info from the METS file. From the METS file I have the ID of the article, and some other metadata, such as the title of the article and the type of the article (which can be article, advertisement, etc). The information highlighted with the green rectangles in the METS file can be linked to the green rectangles from the ALTO files. My goal is to get the following data frame from the METS file:

and this data frame from the ALTO files:

As you can see, by combining both data frames I can know which words come from the same article, which will be helpful for further analysis. A lot of things happened in the 1860s. I am really curious to see if and how these events where reported in a Luxembourguish newspaper. I am particularly curious about how long it took to report certain news from far away, such as the assassination of Abraham Lincoln. But before that I need to extract the data!

I will only focus on the METS file. The logic for the ALTO file is the same. All the source code will be in the appendix of this blog post.

First, let’s take a look at a METS file:

mets <- read_file("1533660_newspaper_lunion_1860-11-14/1533660_newspaper_lunion_1860-11-14-mets.xml")

This is how it looks like:

"<?xml version='1.0' encoding='utf-8'?>\r\n<mets xmlns=\"http://www.loc.gov/METS/\" xmlns:mix=\"http://www.loc.gov/mix/v20\" xmlns:mods=\"http://www.loc.gov/mods/v3\" xmlns:xlink=\"http://www.w3.org/1999/xlink\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" LABEL=\"L'UNION. 1860-11-14_01\" OBJID=\"https://persist.lu/ark:/70795/m62fcm\" TYPE=\"Newspaper\" xsi:schemaLocation=\"http://www.loc.gov/METS/ http://www.loc.gov/standards/mets/mets.xsd http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-6.xsd http://www.loc.gov/mix/v20 http://www.loc.gov/standards/mix/mix.xsd\">\r\n  <metsHdr CREATEDATE=\"2010-12-03T20:35:05\" LASTMODDATE=\"2018-05-09T05:35:51Z\">\r\n    <agent OTHERTYPE=\"SOFTWARE\" ROLE=\"CREATOR\" TYPE=\"OTHER\">\r\n      <name>CCS docWORKS/METAe Version 6.4-3</name>\r\n      <note>docWORKS-ID: 101636</note>\r\n    </agent>\r\n  </metsHdr>\r\n  <dmdSec ID=\"MODSMD_COLLECTION\">\r\n    <mdWrap LABEL=\"Bibliographic meta-data of the collection\" MDTYPE=\"MODS\" MIMETYPE=\"text/xml\">\r\n      <xmlData>\r\n        <mods:mods>\r\n          <mods:identifier type=\"local\">lunion</mods:identifier>\r\n          <mods:titleInfo ID=\"MODSMD_COLLECTION_TI1\" xml:lang=\"fr\">\r\n            <mods:title>L'UNION.</mods:title>\r\n          </mods:titleInfo>\r\n        </mods:mods>\r\n      </xmlData>\r\n    </mdWrap>\r\n  </dmdSec>\r\n  <dmdSec ID=\"MODSMD_SECTION1\">\r\n    <mdWrap MDTYPE=\"MODS\" MIMETYPE=\"text/xml\">\r\n      <xmlData>\r\n        <mods:mods>\r\n          <mods:titleInfo ID=\"MODSMD_SECTION1_TI1\" xml:lang=\"fr\">\r\n            <mods:title>Chemins de fer. — Service d'hiver.</mods:title>\r\n          </mods:titleInfo>\r\n          <mods:language>\r\n            <mods:languageTerm authority=\"rfc3066\" type=\"code\">fr</mods:languageTerm>\r\n ...."

As usual when you import text files like this, it’s always a good idea to split the file. I will split at the "DMDID" character. Take a look back at the second screenshot. The very first tag, first row, first word after div is "DMDID". By splitting at this level, I will get back a list, where each element is the content of this div DMDID block. This is exactly what I need, since this block contains the information from the green rectangles. So let’s split the mets variable at this level:

mets_articles <- mets %>%
    str_split("DMDID") %>%

Let’s take a look at mets_articles:

 chr [1:25] "<?xml version='1.0' encoding='utf-8'?>\r\n<mets xmlns=\"http://www.loc.gov/METS/\" xmlns:mix=\"http://www.loc.g"| __truncated__ ...

Doesn’t seem to be very helpful, but actually it is. We can see that mets_articles is a now a list of 25 elements.

This means that for each element of mets_articles, I need to get the identifier, the label, the type (the red rectangles from the screenshot), but also the information from the "BEGIN" element (the green rectangle).

To do this, I’ll be using regular expressions. In general, I start by experimenting in the console, and then when things start looking good, I write a function. Here is this function:

extractor <- function(string, regex, all = FALSE){
    if(all) {
        string %>%
            str_extract_all(regex) %>%
            flatten_chr() %>%
            str_extract_all("[:alnum:]+", simplify = FALSE) %>%
            map(paste, collapse = "_") %>%
    } else {
        string %>%
            str_extract(regex) %>%
            str_extract_all("[:alnum:]+", simplify = TRUE) %>%
            paste(collapse = " ") %>%

This function may seem complicated, but it simply encapsulates some pretty standard steps to get the data I need. I had to consider two cases. The first case is when I need to extract all the elements with str_extract_all(), or only the first occurrence, with str_extract(). Let’s test it on the first article of the mets_articles list:

mets_articles_1 <- mets_articles[1]
extractor(mets_articles_1, "ID", all = FALSE)
## [1] "id"

Let’s see what happens with all = TRUE:

extractor(mets_articles_1, "ID", all = TRUE)
##   [1] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [15] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [29] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [43] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [57] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [71] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [85] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
##  [99] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
## [113] "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID" "ID"
## [127] "ID" "ID" "ID" "ID" "ID"

This seems to work as intended. Since I need to call this function several times, I’ll be writing another function that extracts all I need:

extract_mets <- function(article){

    id <- article %>%

    label <- article %>%

    type <- article %>%

    begins <- article %>%
        extractor("(?<=BEGIN)(.*?)(?=BETYPE)", all = TRUE)

    tibble::tribble(~label, ~type, ~begins, ~id,
                    label, type, begins, id) %>%

This function uses complex regular expressions to extract the strings I need, and then puts the result into a data frame, with the tibble() function. I then use unnest(), because label, type, begins and id are not the same length. label, type and id are of length 1, while begins is longer. This means that when I put them into a data frame it looks like this:

tribble(~a, ~b,
"a", rep("b", 4))
## # A tibble: 1 x 2
##   a     b        
##   <chr> <list>   
## 1 a     <chr [4]>

With unnest(), I get a nice data frame:

tribble(~a, ~b,
"a", rep("b", 4)) %>% 
## # A tibble: 4 x 2
##   a     b    
##   <chr> <chr>
## 1 a     b    
## 2 a     b    
## 3 a     b    
## 4 a     b

Now, I simply need to map this function to all the files and that’s it! For this, I will write yet another helper function:

mets_csv <- function(page_path){
    page <- read_file(page_path)
    doc_name <- str_extract(page_path, "(?<=/).*")
    mets_articles <- page %>%
        str_split("DMDID") %>%
    mets_df <- map_df(mets_articles, extract_mets)
    mets_df <- mets_df %>%
        mutate(document = doc_name)
    write_csv(mets_df, paste0(page_path, ".csv"))

This function takes the path to a METS file as input, and processes it using the steps I explained above. The only difference is that I add a column containing the name of the file that was processed, and write the resulting data frame directly to disk as a data frame. Finally, I can map this function to all the METS files:

# Extract content from METS files

pages_mets <- str_match(list.files(path = "./", all.files = TRUE, recursive = TRUE), ".*mets.xml") %>%


plan(multiprocess, workers = 8)

tic <- Sys.time()
future_map(pages_mets, mets_csv)
toc <- Sys.time()

toc - tic

I use {furrr} to extract the data from all the files in parallel, by putting 8 cores of my CPU to work. This took around 3 minutes and 20 seconds to finish.

That’s it for now, stay tuned for part 2 where I will analyze this fresh data!

Hope you enjoyed! If you found this blog post useful, you might want to follow me on twitter for blog post updates and buy me an espresso or paypal.me.

Buy me an EspressoBuy me an Espresso


extract_alto <- function(article){
    begins <- article[1] %>%
        extractor("(?<=^ID)(.*?)(?=HPOS)", all = TRUE)

    content <- article %>%
        extractor("(?<=CONTENT)(.*?)(?=WC)", all = TRUE)

    tibble::tribble(~begins, ~content,
                    begins, content) %>%

alto_csv <- function(page_path){

    page <- read_file(page_path)

    doc_name <- str_extract(page_path, "(?<=/text/).*")

    alto_articles <- page %>%
        str_split("TextBlock ") %>%

    alto_df <- map_df(alto_articles, extract_alto)

    alto_df <- alto_df %>%
        mutate(document = doc_name)

    write_csv(alto_df, paste0(page_path, ".csv"))

alto <- read_file("1533660_newspaper_lunion_1860-11-14/text/1860-11-14_01-00001.xml")

# Extract content from alto files

pages_alto <- str_match(list.files(path = "./", all.files = TRUE, recursive = TRUE), ".*/text/.*.xml") %>%


plan(multiprocess, workers = 8)

tic <- Sys.time()
future_map(pages_alto, alto_csv)
toc <- Sys.time()

toc - tic

#Time difference of 18.64776 mins